Gitiles
Code Review Sign In
gerrit-public.fairphone.software / fp2-dev / platform / system / media / ca25627fc63255d147286ef9a9624059825eb548 / . / camera / docs / metadata_properties.xml
blob: fe96508a10d2033670b885f4c3cac1a170dcba33 [file] [log] [blame]
<?xml version="1.0" encoding="utf-8"?>
<!-- Copyright (C) 2012 The Android Open Source Project
Licensed under the Apache License, Version 2.0 (the "License");
you may not use this file except in compliance with the License.
You may obtain a copy of the License at
http://www.apache.org/licenses/LICENSE-2.0
Unless required by applicable law or agreed to in writing, software
distributed under the License is distributed on an "AS IS" BASIS,
WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
See the License for the specific language governing permissions and
limitations under the License.
-->
<metadata xmlns="http://schemas.android.com/service/camera/metadata/"
xmlns:xsi="http://www.w3.org/2001/XMLSchema-instance"
xsi:schemaLocation="http://schemas.android.com/service/camera/metadata/ metadata_properties.xsd">
<tags>
<tag id="BC">
Needed for backwards compatibility with old Java API
</tag>
<tag id="V1">
New features for first camera 2 release (API1)
</tag>
<tag id="RAW">
Needed for useful RAW image processing and DNG file support
</tag>
<tag id="HAL2">
Entry is only used by camera device HAL 2.x
</tag>
<tag id="FULL">
Entry is required for full hardware level devices, and optional for other hardware levels
</tag>
<tag id="FUTURE">
Entry is under-specified and is not required for now. This is for book-keeping purpose,
do not implement or use it, it may be revised for future.
</tag>
</tags>
<types>
<typedef name="pairFloatFloat">
<language name="java">android.util.Pair&lt;Float,Float&gt;</language>
</typedef>
<typedef name="pairDoubleDouble">
<language name="java">android.util.Pair&lt;Double,Double&gt;</language>
</typedef>
<typedef name="rectangle">
<language name="java">android.graphics.Rect</language>
</typedef>
<typedef name="size">
<language name="java">android.util.Size</language>
</typedef>
<typedef name="string">
<language name="java">String</language>
</typedef>
<typedef name="boolean">
<language name="java">boolean</language>
</typedef>
<typedef name="imageFormat">
<language name="java">int</language>
</typedef>
<typedef name="streamConfigurationMap">
<language name="java">android.hardware.camera2.params.StreamConfigurationMap</language>
</typedef>
<typedef name="streamConfiguration">
<language name="java">android.hardware.camera2.params.StreamConfiguration</language>
</typedef>
<typedef name="streamConfigurationDuration">
<language name="java">android.hardware.camera2.params.StreamConfigurationDuration</language>
</typedef>
<typedef name="face">
<language name="java">android.hardware.camera2.params.Face</language>
</typedef>
<typedef name="meteringRectangle">
<language name="java">android.hardware.camera2.params.MeteringRectangle</language>
</typedef>
<typedef name="rangeFloat">
<language name="java">android.util.Range&lt;Float&gt;</language>
</typedef>
<typedef name="rangeInt">
<language name="java">android.util.Range&lt;Integer&gt;</language>
</typedef>
<typedef name="rangeLong">
<language name="java">android.util.Range&lt;Long&gt;</language>
</typedef>
<typedef name="colorSpaceTransform">
<language name="java">android.hardware.camera2.params.ColorSpaceTransform</language>
</typedef>
<typedef name="rggbChannelVector">
<language name="java">android.hardware.camera2.params.RggbChannelVector</language>
</typedef>
<typedef name="blackLevelPattern">
<language name="java">android.hardware.camera2.params.BlackLevelPattern</language>
</typedef>
<typedef name="enumList">
<language name="java">int</language>
</typedef>
<typedef name="sizeF">
<language name="java">android.util.SizeF</language>
</typedef>
<typedef name="point">
<language name="java">android.graphics.Point</language>
</typedef>
<typedef name="tonemapCurve">
<language name="java">android.hardware.camera2.params.TonemapCurve</language>
</typedef>
<typedef name="lensShadingMap">
<language name="java">android.hardware.camera2.params.LensShadingMap</language>
</typedef>
<typedef name="location">
<language name="java">android.location.Location</language>
</typedef>
<typedef name="highSpeedVideoConfiguration">
<language name="java">android.hardware.camera2.params.HighSpeedVideoConfiguration</language>
</typedef>
</types>
<namespace name="android">
<section name="colorCorrection">
<controls>
<entry name="mode" type="byte" visibility="public" enum="true" hwlevel="full">
<enum>
<value>TRANSFORM_MATRIX
<notes>Use the android.colorCorrection.transform matrix
and android.colorCorrection.gains to do color conversion.
All advanced white balance adjustments (not specified
by our white balance pipeline) must be disabled.
If AWB is enabled with `android.control.awbMode != OFF`, then
TRANSFORM_MATRIX is ignored. The camera device will override
this value to either FAST or HIGH_QUALITY.
</notes>
</value>
<value>FAST
<notes>Color correction processing must not slow down
capture rate relative to sensor raw output.
Advanced white balance adjustments above and beyond
the specified white balance pipeline may be applied.
If AWB is enabled with `android.control.awbMode != OFF`, then
the camera device uses the last frame's AWB values
(or defaults if AWB has never been run).
</notes>
</value>
<value>HIGH_QUALITY
<notes>Color correction processing operates at improved
quality but reduced capture rate (relative to sensor raw
output).
Advanced white balance adjustments above and beyond
the specified white balance pipeline may be applied.
If AWB is enabled with `android.control.awbMode != OFF`, then
the camera device uses the last frame's AWB values
(or defaults if AWB has never been run).
</notes>
</value>
</enum>
<description>
The mode control selects how the image data is converted from the
sensor's native color into linear sRGB color.
</description>
<details>
When auto-white balance (AWB) is enabled with android.control.awbMode, this
control is overridden by the AWB routine. When AWB is disabled, the
application controls how the color mapping is performed.
We define the expected processing pipeline below. For consistency
across devices, this is always the case with TRANSFORM_MATRIX.
When either FULL or HIGH_QUALITY is used, the camera device may
do additional processing but android.colorCorrection.gains and
android.colorCorrection.transform will still be provided by the
camera device (in the results) and be roughly correct.
Switching to TRANSFORM_MATRIX and using the data provided from
FAST or HIGH_QUALITY will yield a picture with the same white point
as what was produced by the camera device in the earlier frame.
The expected processing pipeline is as follows:
![White balance processing pipeline](android.colorCorrection.mode/processing_pipeline.png)
The white balance is encoded by two values, a 4-channel white-balance
gain vector (applied in the Bayer domain), and a 3x3 color transform
matrix (applied after demosaic).
The 4-channel white-balance gains are defined as:
android.colorCorrection.gains = [ R G_even G_odd B ]
where `G_even` is the gain for green pixels on even rows of the
output, and `G_odd` is the gain for green pixels on the odd rows.
These may be identical for a given camera device implementation; if
the camera device does not support a separate gain for even/odd green
channels, it will use the `G_even` value, and write `G_odd` equal to
`G_even` in the output result metadata.
The matrices for color transforms are defined as a 9-entry vector:
android.colorCorrection.transform = [ I0 I1 I2 I3 I4 I5 I6 I7 I8 ]
which define a transform from input sensor colors, `P_in = [ r g b ]`,
to output linear sRGB, `P_out = [ r' g' b' ]`,
with colors as follows:
r' = I0r + I1g + I2b
g' = I3r + I4g + I5b
b' = I6r + I7g + I8b
Both the input and output value ranges must match. Overflow/underflow
values are clipped to fit within the range.
</details>
</entry>
<entry name="transform" type="rational" visibility="public"
type_notes="3x3 rational matrix in row-major order"
container="array" typedef="colorSpaceTransform" hwlevel="full">
<array>
<size>3</size>
<size>3</size>
</array>
<description>A color transform matrix to use to transform
from sensor RGB color space to output linear sRGB color space.
</description>
<details>This matrix is either set by the camera device when the request
android.colorCorrection.mode is not TRANSFORM_MATRIX, or
directly by the application in the request when the
android.colorCorrection.mode is TRANSFORM_MATRIX.
In the latter case, the camera device may round the matrix to account
for precision issues; the final rounded matrix should be reported back
in this matrix result metadata. The transform should keep the magnitude
of the output color values within `[0, 1.0]` (assuming input color
values is within the normalized range `[0, 1.0]`), or clipping may occur.
</details>
</entry>
<entry name="gains" type="float" visibility="public"
type_notes="A 1D array of floats for 4 color channel gains"
container="array" typedef="rggbChannelVector" hwlevel="full">
<array>
<size>4</size>
</array>
<description>Gains applying to Bayer raw color channels for
white-balance.</description>
<details>
These per-channel gains are either set by the camera device
when the request android.colorCorrection.mode is not
TRANSFORM_MATRIX, or directly by the application in the
request when the android.colorCorrection.mode is
TRANSFORM_MATRIX.
The gains in the result metadata are the gains actually
applied by the camera device to the current frame.
</details>
<hal_details>
The 4-channel white-balance gains are defined in
the order of `[R G_even G_odd B]`, where `G_even` is the gain
for green pixels on even rows of the output, and `G_odd`
is the gain for green pixels on the odd rows.
If a HAL does not support a separate gain for even/odd green
channels, it must use the `G_even` value, and write
`G_odd` equal to `G_even` in the output result metadata.
</hal_details>
</entry>
<entry name="aberrationMode" type="byte" visibility="public" enum="true" hwlevel="legacy">
<enum>
<value>OFF
<notes>
No aberration correction is applied.
</notes>
</value>
<value>FAST
<notes>
Aberration correction will not slow down capture rate
relative to sensor raw output.
</notes>
</value>
<value>HIGH_QUALITY
<notes>
Aberration correction operates at improved quality but reduced
capture rate (relative to sensor raw output).
</notes>
</value>
</enum>
<description>
Mode of operation for the chromatic aberration correction algorithm.
</description>
<details>
This must be set to a valid mode from
android.colorCorrection.availableAberrationModes.
Chromatic (color) aberration is caused by the fact that different wavelengths of light
can not focus on the same point after exiting from the lens. This metadata defines
the high level control of chromatic aberration correction algorithm, which aims to
minimize the chromatic artifacts that may occur along the object boundaries in an
image.
FAST/HIGH_QUALITY both mean that camera device determined aberration
correction will be applied. HIGH_QUALITY mode indicates that the camera device will
use the highest-quality aberration correction algorithms, even if it slows down
capture rate. FAST means the camera device will not slow down capture rate when
applying aberration correction.
LEGACY devices will always be in FAST mode.
</details>
</entry>
</controls>
<dynamic>
<clone entry="android.colorCorrection.mode" kind="controls">
</clone>
<clone entry="android.colorCorrection.transform" kind="controls">
</clone>
<clone entry="android.colorCorrection.gains" kind="controls">
</clone>
<clone entry="android.colorCorrection.aberrationMode" kind="controls">
</clone>
</dynamic>
<static>
<entry name="availableAberrationModes" type="byte" visibility="public"
type_notes="list of enums" container="array" typedef="enumList" hwlevel="legacy">
<array>
<size>n</size>
</array>
<description>
The set of aberration correction modes supported by this camera device.
</description>
<range>
The subset of aberration correction mode from those specified in
android.colorCorrection.aberrationMode.
</range>
<details>
This metadata lists the valid modes for android.colorCorrection.aberrationMode.
If no aberration correction modes are available for a device, this list will solely include
OFF mode.
For FULL capability device (android.info.supportedHardwareLevel `==` FULL), OFF must be
included.
LEGACY devices will always only support FAST mode.
</details>
<tag id="V1" />
</entry>
</static>
</section>
<section name="control">
<controls>
<entry name="aeAntibandingMode" type="byte" visibility="public"
enum="true" hwlevel="legacy">
<enum>
<value>OFF
<notes>
The camera device will not adjust exposure duration to
avoid banding problems.
</notes>
</value>
<value>50HZ
<notes>
The camera device will adjust exposure duration to
avoid banding problems with 50Hz illumination sources.
</notes>
</value>
<value>60HZ
<notes>
The camera device will adjust exposure duration to
avoid banding problems with 60Hz illumination
sources.
</notes>
</value>
<value>AUTO
<notes>
The camera device will automatically adapt its
antibanding routine to the current illumination
conditions. This is the default.
</notes>
</value>
</enum>
<description>
The desired setting for the camera device's auto-exposure
algorithm's antibanding compensation.
</description>
<range>
android.control.aeAvailableAntibandingModes
</range>
<details>
Some kinds of lighting fixtures, such as some fluorescent
lights, flicker at the rate of the power supply frequency
(60Hz or 50Hz, depending on country). While this is
typically not noticeable to a person, it can be visible to
a camera device. If a camera sets its exposure time to the
wrong value, the flicker may become visible in the
viewfinder as flicker or in a final captured image, as a
set of variable-brightness bands across the image.
Therefore, the auto-exposure routines of camera devices
include antibanding routines that ensure that the chosen
exposure value will not cause such banding. The choice of
exposure time depends on the rate of flicker, which the
camera device can detect automatically, or the expected
rate can be selected by the application using this
control.
A given camera device may not support all of the possible
options for the antibanding mode. The
android.control.aeAvailableAntibandingModes key contains
the available modes for a given camera device.
The default mode is AUTO, which must be supported by all
camera devices.
If manual exposure control is enabled (by setting
android.control.aeMode or android.control.mode to OFF),
then this setting has no effect, and the application must
ensure it selects exposure times that do not cause banding
issues. The android.statistics.sceneFlicker key can assist
the application in this.
</details>
<hal_details>
For all capture request templates, this field must be set
to AUTO. AUTO is the only mode that must supported;
OFF, 50HZ, 60HZ are all optional.
If manual exposure control is enabled (by setting
android.control.aeMode or android.control.mode to OFF),
then the exposure values provided by the application must not be
adjusted for antibanding.
</hal_details>
<tag id="BC" />
</entry>
<entry name="aeExposureCompensation" type="int32" visibility="public" hwlevel="legacy">
<description>Adjustment to auto-exposure (AE) target image
brightness.</description>
<units>count of positive/negative EV steps</units>
<range>android.control.aeCompensationRange</range>
<details>
The adjustment is measured as a count of steps, with the
step size defined by android.control.aeCompensationStep and the
allowed range by android.control.aeCompensationRange.
For example, if the exposure value (EV) step is 0.333, '6'
will mean an exposure compensation of +2 EV; -3 will mean an
exposure compensation of -1 EV. One EV represents a doubling
of image brightness. Note that this control will only be
effective if android.control.aeMode `!=` OFF. This control
will take effect even when android.control.aeLock `== true`.
In the event of exposure compensation value being changed, camera device
may take several frames to reach the newly requested exposure target.
During that time, android.control.aeState field will be in the SEARCHING
state. Once the new exposure target is reached, android.control.aeState will
change from SEARCHING to either CONVERGED, LOCKED (if AE lock is enabled), or
FLASH_REQUIRED (if the scene is too dark for still capture).
</details>
<tag id="BC" />
</entry>
<entry name="aeLock" type="byte" visibility="public" enum="true"
typedef="boolean" hwlevel="legacy">
<enum>
<value>OFF
<notes>Auto-exposure lock is disabled; the AE algorithm
is free to update its parameters.</notes></value>
<value>ON
<notes>Auto-exposure lock is enabled; the AE algorithm
must not update the exposure and sensitivity parameters
while the lock is active.
android.control.aeExposureCompensation setting changes
will still take effect while auto-exposure is locked.
Some rare LEGACY devices may not support
this, in which case the value will always be overridden to OFF.
</notes></value>
</enum>
<description>Whether auto-exposure (AE) is currently locked to its latest
calculated values.</description>
<details>Note that even when AE is locked, the flash may be
fired if the android.control.aeMode is ON_AUTO_FLASH / ON_ALWAYS_FLASH /
ON_AUTO_FLASH_REDEYE.
When android.control.aeExposureCompensation is changed, even if the AE lock
is ON, the camera device will still adjust its exposure value.
If AE precapture is triggered (see android.control.aePrecaptureTrigger)
when AE is already locked, the camera device will not change the exposure time
(android.sensor.exposureTime) and sensitivity (android.sensor.sensitivity)
parameters. The flash may be fired if the android.control.aeMode
is ON_AUTO_FLASH/ON_AUTO_FLASH_REDEYE and the scene is too dark. If the
android.control.aeMode is ON_ALWAYS_FLASH, the scene may become overexposed.
See android.control.aeState for AE lock related state transition details.
</details>
<tag id="BC" />
</entry>
<entry name="aeMode" type="byte" visibility="public" enum="true" hwlevel="legacy">
<enum>
<value>OFF
<notes>
The camera device's autoexposure routine is disabled.
The application-selected android.sensor.exposureTime,
android.sensor.sensitivity and
android.sensor.frameDuration are used by the camera
device, along with android.flash.* fields, if there's
a flash unit for this camera device.
LEGACY devices do not support the OFF mode and will
override attempts to use this value to ON.
</notes>
</value>
<value>ON
<notes>
The camera device's autoexposure routine is active,
with no flash control.
The application's values for
android.sensor.exposureTime,
android.sensor.sensitivity, and
android.sensor.frameDuration are ignored. The
application has control over the various
android.flash.* fields.
</notes>
</value>
<value>ON_AUTO_FLASH
<notes>
Like ON, except that the camera device also controls
the camera's flash unit, firing it in low-light
conditions.
The flash may be fired during a precapture sequence
(triggered by android.control.aePrecaptureTrigger) and
may be fired for captures for which the
android.control.captureIntent field is set to
STILL_CAPTURE
</notes>
</value>
<value>ON_ALWAYS_FLASH
<notes>
Like ON, except that the camera device also controls
the camera's flash unit, always firing it for still
captures.
The flash may be fired during a precapture sequence
(triggered by android.control.aePrecaptureTrigger) and
will always be fired for captures for which the
android.control.captureIntent field is set to
STILL_CAPTURE
</notes>
</value>
<value>ON_AUTO_FLASH_REDEYE
<notes>
Like ON_AUTO_FLASH, but with automatic red eye
reduction.
If deemed necessary by the camera device, a red eye
reduction flash will fire during the precapture
sequence.
</notes>
</value>
</enum>
<description>The desired mode for the camera device's
auto-exposure routine.</description>
<range>android.control.aeAvailableModes</range>
<details>
This control is only effective if android.control.mode is
AUTO.
When set to any of the ON modes, the camera device's
auto-exposure routine is enabled, overriding the
application's selected exposure time, sensor sensitivity,
and frame duration (android.sensor.exposureTime,
android.sensor.sensitivity, and
android.sensor.frameDuration). If one of the FLASH modes
is selected, the camera device's flash unit controls are
also overridden.
The FLASH modes are only available if the camera device
has a flash unit (android.flash.info.available is `true`).
If flash TORCH mode is desired, this field must be set to
ON or OFF, and android.flash.mode set to TORCH.
When set to any of the ON modes, the values chosen by the
camera device auto-exposure routine for the overridden
fields for a given capture will be available in its
CaptureResult.
</details>
<tag id="BC" />
</entry>
<entry name="aeRegions" type="int32" visibility="public"
optional="true" container="array" typedef="meteringRectangle">
<array>
<size>5</size>
<size>area_count</size>
</array>
<description>List of areas to use for
metering.</description>
<range>`area_count &lt;= android.control.maxRegionsAe`</range>
<details>
Optional. Not available if android.control.maxRegionsAe is 0.
Otherwise will always be present.
The coordinate system is based on the active pixel array,
with (0,0) being the top-left pixel in the active pixel array, and
(android.sensor.info.activeArraySize.width - 1,
android.sensor.info.activeArraySize.height - 1) being the
bottom-right pixel in the active pixel array.
The weight must range from 0 to 1000, and represents a weight
for every pixel in the area. This means that a large metering area
with the same weight as a smaller area will have more effect in
the metering result. Metering areas can partially overlap and the
camera device will add the weights in the overlap region.
If all regions have 0 weight, then no specific metering area
needs to be used by the camera device. If the metering region is
outside the used android.scaler.cropRegion returned in capture result metadata,
the camera device will ignore the sections outside the region and output the
used sections in the result metadata.
</details>
<hal_details>
The HAL level representation of MeteringRectangle[] is a
int[5 * area_count].
Every five elements represent a metering region of
(xmin, ymin, xmax, ymax, weight).
The rectangle is defined to be inclusive on xmin and ymin, but
exclusive on xmax and ymax.
</hal_details>
<tag id="BC" />
</entry>
<entry name="aeTargetFpsRange" type="int32" visibility="public"
container="array" typedef="rangeInt" hwlevel="legacy">
<array>
<size>2</size>
</array>
<description>Range over which fps can be adjusted to
maintain exposure.</description>
<range>android.control.aeAvailableTargetFpsRanges</range>
<details>Only constrains auto-exposure (AE) algorithm, not
manual control of android.sensor.exposureTime</details>
<tag id="BC" />
</entry>
<entry name="aePrecaptureTrigger" type="byte" visibility="public"
enum="true" hwlevel="limited">
<enum>
<value>IDLE
<notes>The trigger is idle.</notes>
</value>
<value>START
<notes>The precapture metering sequence will be started
by the camera device.
The exact effect of the precapture trigger depends on
the current AE mode and state.</notes>
</value>
</enum>
<description>Whether the camera device will trigger a precapture
metering sequence when it processes this request.</description>
<details>This entry is normally set to IDLE, or is not
included at all in the request settings. When included and
set to START, the camera device will trigger the autoexposure
precapture metering sequence.
The precapture sequence should triggered before starting a
high-quality still capture for final metering decisions to
be made, and for firing pre-capture flash pulses to estimate
scene brightness and required final capture flash power, when
the flash is enabled.
Normally, this entry should be set to START for only a
single request, and the application should wait until the
sequence completes before starting a new one.
The exact effect of auto-exposure (AE) precapture trigger
depends on the current AE mode and state; see
android.control.aeState for AE precapture state transition
details.</details>
<tag id="BC" />
</entry>
<entry name="afMode" type="byte" visibility="public" enum="true"
hwlevel="legacy">
<enum>
<value>OFF
<notes>The auto-focus routine does not control the lens;
android.lens.focusDistance is controlled by the
application.</notes></value>
<value>AUTO
<notes>Basic automatic focus mode.
In this mode, the lens does not move unless
the autofocus trigger action is called. When that trigger
is activated, AF will transition to ACTIVE_SCAN, then to
the outcome of the scan (FOCUSED or NOT_FOCUSED).
Always supported if lens is not fixed focus.
Use android.lens.info.minimumFocusDistance to determine if lens
is fixed-focus.
Triggering AF_CANCEL resets the lens position to default,
and sets the AF state to INACTIVE.</notes></value>
<value>MACRO
<notes>Close-up focusing mode.
In this mode, the lens does not move unless the
autofocus trigger action is called. When that trigger is
activated, AF will transition to ACTIVE_SCAN, then to
the outcome of the scan (FOCUSED or NOT_FOCUSED). This
mode is optimized for focusing on objects very close to
the camera.
When that trigger is activated, AF will transition to
ACTIVE_SCAN, then to the outcome of the scan (FOCUSED or
NOT_FOCUSED). Triggering cancel AF resets the lens
position to default, and sets the AF state to
INACTIVE.</notes></value>
<value>CONTINUOUS_VIDEO
<notes>In this mode, the AF algorithm modifies the lens
position continually to attempt to provide a
constantly-in-focus image stream.
The focusing behavior should be suitable for good quality
video recording; typically this means slower focus
movement and no overshoots. When the AF trigger is not
involved, the AF algorithm should start in INACTIVE state,
and then transition into PASSIVE_SCAN and PASSIVE_FOCUSED
states as appropriate. When the AF trigger is activated,
the algorithm should immediately transition into
AF_FOCUSED or AF_NOT_FOCUSED as appropriate, and lock the
lens position until a cancel AF trigger is received.
Once cancel is received, the algorithm should transition
back to INACTIVE and resume passive scan. Note that this
behavior is not identical to CONTINUOUS_PICTURE, since an
ongoing PASSIVE_SCAN must immediately be
canceled.</notes></value>
<value>CONTINUOUS_PICTURE
<notes>In this mode, the AF algorithm modifies the lens
position continually to attempt to provide a
constantly-in-focus image stream.
The focusing behavior should be suitable for still image
capture; typically this means focusing as fast as
possible. When the AF trigger is not involved, the AF
algorithm should start in INACTIVE state, and then
transition into PASSIVE_SCAN and PASSIVE_FOCUSED states as
appropriate as it attempts to maintain focus. When the AF
trigger is activated, the algorithm should finish its
PASSIVE_SCAN if active, and then transition into
AF_FOCUSED or AF_NOT_FOCUSED as appropriate, and lock the
lens position until a cancel AF trigger is received.
When the AF cancel trigger is activated, the algorithm
should transition back to INACTIVE and then act as if it
has just been started.</notes></value>
<value>EDOF
<notes>Extended depth of field (digital focus) mode.
The camera device will produce images with an extended
depth of field automatically; no special focusing
operations need to be done before taking a picture.
AF triggers are ignored, and the AF state will always be
INACTIVE.</notes></value>
</enum>
<description>Whether auto-focus (AF) is currently enabled, and what
mode it is set to.</description>
<range>android.control.afAvailableModes</range>
<details>Only effective if android.control.mode = AUTO and the lens is not fixed focus
(i.e. `android.lens.info.minimumFocusDistance &gt; 0`).
If the lens is controlled by the camera device auto-focus algorithm,
the camera device will report the current AF status in android.control.afState
in result metadata.</details>
<hal_details>
When afMode is AUTO or MACRO, the lens must not move until an AF trigger is sent in a
request (android.control.afTrigger `==` START). After an AF trigger, the afState will end
up with either FOCUSED_LOCKED or NOT_FOCUSED_LOCKED state (see
android.control.afState for detailed state transitions), which indicates that the lens is
locked and will not move. If camera movement (e.g. tilting camera) causes the lens to move
after the lens is locked, the HAL must compensate this movement appropriately such that
the same focal plane remains in focus.
When afMode is one of the continuous auto focus modes, the HAL is free to start a AF
scan whenever it's not locked. When the lens is locked after an AF trigger
(see android.control.afState for detailed state transitions), the HAL should maintain the
same lock behavior as above.
When afMode is OFF, the application controls focus manually. The accuracy of the
focus distance control depends on the android.lens.info.focusDistanceCalibration.
However, the lens must not move regardless of the camera movement for any focus distance
manual control.
To put this in concrete terms, if the camera has lens elements which may move based on
camera orientation or motion (e.g. due to gravity), then the HAL must drive the lens to
remain in a fixed position invariant to the camera's orientation or motion, for example,
by using accelerometer measurements in the lens control logic. This is a typical issue
that will arise on camera modules with open-loop VCMs.
</hal_details>
<tag id="BC" />
</entry>
<entry name="afRegions" type="int32" visibility="public"
optional="true" container="array" typedef="meteringRectangle">
<array>
<size>5</size>
<size>area_count</size>
</array>
<description>List of areas to use for focus
estimation.</description>
<range>`area_count &lt;= android.control.maxRegions[2]`</range>
<details>
Optional. Not available if android.control.maxRegionsAf is 0.
Otherwise will always be present.
The coordinate system is based on the active pixel array,
with (0,0) being the top-left pixel in the active pixel array, and
(android.sensor.info.activeArraySize.width - 1,
android.sensor.info.activeArraySize.height - 1) being the
bottom-right pixel in the active pixel array.
The weight must range from 0 to 1000, and represents a weight
for every pixel in the area. This means that a large metering area
with the same weight as a smaller area will have more effect in
the metering result. Metering areas can partially overlap and the
camera device will add the weights in the overlap region.
If all regions have 0 weight, then no specific metering area
needs to be used by the camera device. If the metering region is
outside the used android.scaler.cropRegion returned in capture result metadata,
the camera device will ignore the sections outside the region and output the
used sections in the result metadata.
</details>
<hal_details>
The HAL level representation of MeteringRectangle[] is a
int[5 * area_count].
Every five elements represent a metering region of
(xmin, ymin, xmax, ymax, weight).
The rectangle is defined to be inclusive on xmin and ymin, but
exclusive on xmax and ymax.
</hal_details>
<tag id="BC" />
</entry>
<entry name="afTrigger" type="byte" visibility="public" enum="true"
hwlevel="legacy">
<enum>
<value>IDLE
<notes>The trigger is idle.</notes>
</value>
<value>START
<notes>Autofocus will trigger now.</notes>
</value>
<value>CANCEL
<notes>Autofocus will return to its initial
state, and cancel any currently active trigger.</notes>
</value>
</enum>
<description>
Whether the camera device will trigger autofocus for this request.
</description>
<details>This entry is normally set to IDLE, or is not
included at all in the request settings.
When included and set to START, the camera device will trigger the
autofocus algorithm. If autofocus is disabled, this trigger has no effect.
When set to CANCEL, the camera device will cancel any active trigger,
and return to its initial AF state.
Generally, applications should set this entry to START or CANCEL for only a
single capture, and then return it to IDLE (or not set at all). Specifying
START for multiple captures in a row means restarting the AF operation over
and over again.
See android.control.afState for what the trigger means for each AF mode.
</details>
<tag id="BC" />
</entry>
<entry name="awbLock" type="byte" visibility="public" enum="true"
typedef="boolean" hwlevel="legacy">
<enum>
<value>OFF
<notes>Auto-white balance lock is disabled; the AWB
algorithm is free to update its parameters if in AUTO
mode.</notes></value>
<value>ON
<notes>Auto-white balance lock is enabled; the AWB
algorithm will not update its parameters while the lock
is active.</notes></value>
</enum>
<description>Whether auto-white balance (AWB) is currently locked to its
latest calculated values.</description>
<details>Note that AWB lock is only meaningful when
android.control.awbMode is in the AUTO mode; in other modes,
AWB is already fixed to a specific setting.
Some LEGACY devices may not support ON; the value is then overridden to OFF.
</details>
<tag id="BC" />
</entry>
<entry name="awbMode" type="byte" visibility="public" enum="true"
hwlevel="legacy">
<enum>
<value>OFF
<notes>
The camera device's auto-white balance routine is disabled.
The application-selected color transform matrix
(android.colorCorrection.transform) and gains
(android.colorCorrection.gains) are used by the camera
device for manual white balance control.
</notes>
</value>
<value>AUTO
<notes>
The camera device's auto-white balance routine is active.
The application's values for android.colorCorrection.transform
and android.colorCorrection.gains are ignored.
For devices that support the MANUAL_POST_PROCESSING capability, the
values used by the camera device for the transform and gains
will be available in the capture result for this request.
</notes>
</value>
<value>INCANDESCENT
<notes>
The camera device's auto-white balance routine is disabled;
the camera device uses incandescent light as the assumed scene
illumination for white balance.
While the exact white balance transforms are up to the
camera device, they will approximately match the CIE
standard illuminant A.
The application's values for android.colorCorrection.transform
and android.colorCorrection.gains are ignored.
For devices that support the MANUAL_POST_PROCESSING capability, the
values used by the camera device for the transform and gains
will be available in the capture result for this request.
</notes>
</value>
<value>FLUORESCENT
<notes>
The camera device's auto-white balance routine is disabled;
the camera device uses fluorescent light as the assumed scene
illumination for white balance.
While the exact white balance transforms are up to the
camera device, they will approximately match the CIE
standard illuminant F2.
The application's values for android.colorCorrection.transform
and android.colorCorrection.gains are ignored.
For devices that support the MANUAL_POST_PROCESSING capability, the
values used by the camera device for the transform and gains
will be available in the capture result for this request.
</notes>
</value>
<value>WARM_FLUORESCENT
<notes>
The camera device's auto-white balance routine is disabled;
the camera device uses warm fluorescent light as the assumed scene
illumination for white balance.
While the exact white balance transforms are up to the
camera device, they will approximately match the CIE
standard illuminant F4.
The application's values for android.colorCorrection.transform
and android.colorCorrection.gains are ignored.
For devices that support the MANUAL_POST_PROCESSING capability, the
values used by the camera device for the transform and gains
will be available in the capture result for this request.
</notes>
</value>
<value>DAYLIGHT
<notes>
The camera device's auto-white balance routine is disabled;
the camera device uses daylight light as the assumed scene
illumination for white balance.
While the exact white balance transforms are up to the
camera device, they will approximately match the CIE
standard illuminant D65.
The application's values for android.colorCorrection.transform
and android.colorCorrection.gains are ignored.
For devices that support the MANUAL_POST_PROCESSING capability, the
values used by the camera device for the transform and gains
will be available in the capture result for this request.
</notes>
</value>
<value>CLOUDY_DAYLIGHT
<notes>
The camera device's auto-white balance routine is disabled;
the camera device uses cloudy daylight light as the assumed scene
illumination for white balance.
The application's values for android.colorCorrection.transform
and android.colorCorrection.gains are ignored.
For devices that support the MANUAL_POST_PROCESSING capability, the
values used by the camera device for the transform and gains
will be available in the capture result for this request.
</notes>
</value>
<value>TWILIGHT
<notes>
The camera device's auto-white balance routine is disabled;
the camera device uses twilight light as the assumed scene
illumination for white balance.
The application's values for android.colorCorrection.transform
and android.colorCorrection.gains are ignored.
For devices that support the MANUAL_POST_PROCESSING capability, the
values used by the camera device for the transform and gains
will be available in the capture result for this request.
</notes>
</value>
<value>SHADE
<notes>
The camera device's auto-white balance routine is disabled;
the camera device uses shade light as the assumed scene
illumination for white balance.
The application's values for android.colorCorrection.transform
and android.colorCorrection.gains are ignored.
For devices that support the MANUAL_POST_PROCESSING capability, the
values used by the camera device for the transform and gains
will be available in the capture result for this request.
</notes>
</value>
</enum>
<description>Whether auto-white balance (AWB) is currently setting the color
transform fields, and what its illumination target
is.</description>
<range>android.control.awbAvailableModes</range>
<details>
This control is only effective if android.control.mode is AUTO.
When set to the ON mode, the camera device's auto-white balance
routine is enabled, overriding the application's selected
android.colorCorrection.transform, android.colorCorrection.gains and
android.colorCorrection.mode.
When set to the OFF mode, the camera device's auto-white balance
routine is disabled. The application manually controls the white
balance by android.colorCorrection.transform, android.colorCorrection.gains
and android.colorCorrection.mode.
When set to any other modes, the camera device's auto-white
balance routine is disabled. The camera device uses each
particular illumination target for white balance
adjustment. The application's values for
android.colorCorrection.transform,
android.colorCorrection.gains and
android.colorCorrection.mode are ignored.
</details>
<tag id="BC" />
</entry>
<entry name="awbRegions" type="int32" visibility="public"
optional="true" container="array" typedef="meteringRectangle">
<array>
<size>5</size>
<size>area_count</size>
</array>
<description>List of areas to use for illuminant
estimation.</description>
<range>`area_count &lt;= android.control.maxRegions[1]`</range>
<details>
Optional. Not available if android.control.maxRegionsAwb is 0.
Otherwise will always be present.
The coordinate system is based on the active pixel array,
with (0,0) being the top-left pixel in the active pixel array, and
(android.sensor.info.activeArraySize.width - 1,
android.sensor.info.activeArraySize.height - 1) being the
bottom-right pixel in the active pixel array.
The weight must range from 0 to 1000, and represents a weight
for every pixel in the area. This means that a large metering area
with the same weight as a smaller area will have more effect in
the metering result. Metering areas can partially overlap and the
camera device will add the weights in the overlap region.
If all regions have 0 weight, then no specific metering area
needs to be used by the camera device. If the metering region is
outside the used android.scaler.cropRegion returned in capture result metadata,
the camera device will ignore the sections outside the region and output the
used sections in the result metadata.
</details>
<hal_details>
The HAL level representation of MeteringRectangle[] is a
int[5 * area_count].
Every five elements represent a metering region of
(xmin, ymin, xmax, ymax, weight).
The rectangle is defined to be inclusive on xmin and ymin, but
exclusive on xmax and ymax.
</hal_details>
<tag id="BC" />
</entry>
<entry name="captureIntent" type="byte" visibility="public" enum="true"
hwlevel="legacy">
<enum>
<value>CUSTOM
<notes>The goal of this request doesn't fall into the other
categories. The camera device will default to preview-like
behavior.</notes></value>
<value>PREVIEW
<notes>This request is for a preview-like use case.
The precapture trigger may be used to start off a metering
w/flash sequence.
</notes></value>
<value>STILL_CAPTURE
<notes>This request is for a still capture-type
use case.
If the flash unit is under automatic control, it may fire as needed.
</notes></value>
<value>VIDEO_RECORD
<notes>This request is for a video recording
use case.</notes></value>
<value>VIDEO_SNAPSHOT
<notes>This request is for a video snapshot (still
image while recording video) use case.
The camera device should take the highest-quality image
possible (given the other settings) without disrupting the
frame rate of video recording. </notes></value>
<value>ZERO_SHUTTER_LAG
<notes>This request is for a ZSL usecase; the
application will stream full-resolution images and
reprocess one or several later for a final
capture.
</notes></value>
<value>MANUAL
<notes>This request is for manual capture use case where
the applications want to directly control the capture parameters.
For example, the application may wish to manually control
android.sensor.exposureTime, android.sensor.sensitivity, etc.
</notes></value>
</enum>
<description>Information to the camera device 3A (auto-exposure,
auto-focus, auto-white balance) routines about the purpose
of this capture, to help the camera device to decide optimal 3A
strategy.</description>
<range>All must be supported except for ZERO_SHUTTER_LAG and MANUAL.</range>
<details>This control (except for MANUAL) is only effective if
`android.control.mode != OFF` and any 3A routine is active.
ZERO_SHUTTER_LAG will be supported if android.request.availableCapabilities
contains ZSL. MANUAL will be supported if android.request.availableCapabilities
contains MANUAL_SENSOR.</details>
<tag id="BC" />
</entry>
<entry name="effectMode" type="byte" visibility="public" enum="true"
hwlevel="legacy">
<enum>
<value>OFF
<notes>
No color effect will be applied.
</notes>
</value>
<value optional="true">MONO
<notes>
A "monocolor" effect where the image is mapped into
a single color.
This will typically be grayscale.
</notes>
</value>
<value optional="true">NEGATIVE
<notes>
A "photo-negative" effect where the image's colors
are inverted.
</notes>
</value>
<value optional="true">SOLARIZE
<notes>
A "solarisation" effect (Sabattier effect) where the
image is wholly or partially reversed in
tone.
</notes>
</value>
<value optional="true">SEPIA
<notes>
A "sepia" effect where the image is mapped into warm
gray, red, and brown tones.
</notes>
</value>
<value optional="true">POSTERIZE
<notes>
A "posterization" effect where the image uses
discrete regions of tone rather than a continuous
gradient of tones.
</notes>
</value>
<value optional="true">WHITEBOARD
<notes>
A "whiteboard" effect where the image is typically displayed
as regions of white, with black or grey details.
</notes>
</value>
<value optional="true">BLACKBOARD
<notes>
A "blackboard" effect where the image is typically displayed
as regions of black, with white or grey details.
</notes>
</value>
<value optional="true">AQUA
<notes>
An "aqua" effect where a blue hue is added to the image.
</notes>
</value>
</enum>
<description>A special color effect to apply.</description>
<range>android.control.availableEffects</range>
<details>
When this mode is set, a color effect will be applied
to images produced by the camera device. The interpretation
and implementation of these color effects is left to the
implementor of the camera device, and should not be
depended on to be consistent (or present) across all
devices.
A color effect will only be applied if
android.control.mode != OFF.
</details>
<tag id="BC" />
</entry>
<entry name="mode" type="byte" visibility="public" enum="true"
hwlevel="legacy">
<enum>
<value>OFF
<notes>Full application control of pipeline.
All control by the device's metering and focusing (3A)
routines is disabled, and no other settings in
android.control.* have any effect, except that
android.control.captureIntent may be used by the camera
device to select post-processing values for processing
blocks that do not allow for manual control, or are not
exposed by the camera API.
However, the camera device's 3A routines may continue to
collect statistics and update their internal state so that
when control is switched to AUTO mode, good control values
can be immediately applied.
</notes></value>
<value>AUTO
<notes>Use settings for each individual 3A routine.
Manual control of capture parameters is disabled. All
controls in android.control.* besides sceneMode take
effect.</notes></value>
<value>USE_SCENE_MODE
<notes>Use a specific scene mode.
Enabling this disables control.aeMode, control.awbMode and
control.afMode controls; the camera device will ignore
those settings while USE_SCENE_MODE is active (except for
FACE_PRIORITY scene mode). Other control entries are still
active. This setting can only be used if scene mode is
supported (i.e. android.control.availableSceneModes
contain some modes other than DISABLED).</notes></value>
<value>OFF_KEEP_STATE
<notes>Same as OFF mode, except that this capture will not be
used by camera device background auto-exposure, auto-white balance and
auto-focus algorithms (3A) to update their statistics.
Specifically, the 3A routines are locked to the last
values set from a request with AUTO, OFF, or
USE_SCENE_MODE, and any statistics or state updates
collected from manual captures with OFF_KEEP_STATE will be
discarded by the camera device.
</notes></value>
</enum>
<description>Overall mode of 3A control
routines.</description>
<range>all must be supported</range>
<details>High-level 3A control. When set to OFF, all 3A control
by the camera device is disabled. The application must set the fields for
capture parameters itself.
When set to AUTO, the individual algorithm controls in
android.control.* are in effect, such as android.control.afMode.
When set to USE_SCENE_MODE, the individual controls in
android.control.* are mostly disabled, and the camera device implements
one of the scene mode settings (such as ACTION, SUNSET, or PARTY)
as it wishes. The camera device scene mode 3A settings are provided by
android.control.sceneModeOverrides.
When set to OFF_KEEP_STATE, it is similar to OFF mode, the only difference
is that this frame will not be used by camera device background 3A statistics
update, as if this frame is never captured. This mode can be used in the scenario
where the application doesn't want a 3A manual control capture to affect
the subsequent auto 3A capture results.
LEGACY mode devices will only support AUTO and USE_SCENE_MODE modes.
LIMITED mode devices will only support OFF and OFF_KEEP_STATE if they
support the MANUAL_SENSOR capability.
</details>
<tag id="BC" />
</entry>
<entry name="sceneMode" type="byte" visibility="public" enum="true"
hwlevel="legacy">
<enum>
<value id="0">DISABLED
<notes>
Indicates that no scene modes are set for a given capture request.
</notes>
</value>
<value>FACE_PRIORITY
<notes>If face detection support exists, use face
detection data for auto-focus, auto-white balance, and
auto-exposure routines.
If face detection statistics are disabled
(i.e. android.statistics.faceDetectMode is set to OFF),
this should still operate correctly (but will not return
face detection statistics to the framework).
Unlike the other scene modes, android.control.aeMode,
android.control.awbMode, and android.control.afMode
remain active when FACE_PRIORITY is set.
</notes>
</value>
<value optional="true">ACTION
<notes>
Optimized for photos of quickly moving objects.
Similar to SPORTS.
</notes>
</value>
<value optional="true">PORTRAIT
<notes>
Optimized for still photos of people.
</notes>
</value>
<value optional="true">LANDSCAPE
<notes>
Optimized for photos of distant macroscopic objects.
</notes>
</value>
<value optional="true">NIGHT
<notes>
Optimized for low-light settings.
</notes>
</value>
<value optional="true">NIGHT_PORTRAIT
<notes>
Optimized for still photos of people in low-light
settings.
</notes>
</value>
<value optional="true">THEATRE
<notes>
Optimized for dim, indoor settings where flash must
remain off.
</notes>
</value>
<value optional="true">BEACH
<notes>
Optimized for bright, outdoor beach settings.
</notes>
</value>
<value optional="true">SNOW
<notes>
Optimized for bright, outdoor settings containing snow.
</notes>
</value>
<value optional="true">SUNSET
<notes>
Optimized for scenes of the setting sun.
</notes>
</value>
<value optional="true">STEADYPHOTO
<notes>
Optimized to avoid blurry photos due to small amounts of
device motion (for example: due to hand shake).
</notes>
</value>
<value optional="true">FIREWORKS
<notes>
Optimized for nighttime photos of fireworks.
</notes>
</value>
<value optional="true">SPORTS
<notes>
Optimized for photos of quickly moving people.
Similar to ACTION.
</notes>
</value>
<value optional="true">PARTY
<notes>
Optimized for dim, indoor settings with multiple moving
people.
</notes>
</value>
<value optional="true">CANDLELIGHT
<notes>
Optimized for dim settings where the main light source
is a flame.
</notes>
</value>
<value optional="true">BARCODE
<notes>
Optimized for accurately capturing a photo of barcode
for use by camera applications that wish to read the
barcode value.
</notes>
</value>
<value optional="true">HIGH_SPEED_VIDEO
<notes>
Optimized for high speed video recording (frame rate >=60fps) use case.
The supported high speed video sizes and fps ranges are specified in
android.control.availableHighSpeedVideoConfigurations. To get desired
output frame rates, the application is only allowed to select video size
and fps range combinations listed in this static metadata. The fps range
can be control via android.control.aeTargetFpsRange.
In this mode, the camera device will override aeMode, awbMode, and afMode to
ON, ON, and CONTINUOUS_VIDEO, respectively. All post-processing block mode
controls will be overridden to be FAST. Therefore, no manual control of capture
and post-processing parameters is possible. All other controls operate the
same as when android.control.mode == AUTO. This means that all other
android.control.* fields continue to work, such as
* android.control.aeTargetFpsRange
* android.control.aeExposureCompensation
* android.control.aeLock
* android.control.awbLock
* android.control.effectMode
* android.control.aeRegions
* android.control.afRegions
* android.control.awbRegions
* android.control.afTrigger
* android.control.aePrecaptureTrigger
Outside of android.control.*, the following controls will work:
* android.flash.mode (automatic flash for still capture will not work since aeMode is ON)
* android.lens.opticalStabilizationMode (if it is supported)
* android.scaler.cropRegion
* android.statistics.faceDetectMode
For high speed recording use case, the actual maximum supported frame rate may
be lower than what camera can output, depending on the destination Surfaces for
the image data. For example, if the destination surface is from video encoder,
the application need check if the video encoder is capable of supporting the
high frame rate for a given video size, or it will end up with lower recording
frame rate. If the destination surface is from preview window, the preview frame
rate will be bounded by the screen refresh rate.
The camera device will only support up to 2 output high speed streams
(processed non-stalling format defined in android.request.maxNumOutputStreams)
in this mode. This control will be effective only if all of below conditions are true:
* The application created no more than maxNumHighSpeedStreams processed non-stalling
format output streams, where maxNumHighSpeedStreams is calculated as
min(2, android.request.maxNumOutputStreams[Processed (but not-stalling)]).
* The stream sizes are selected from the sizes reported by
android.control.availableHighSpeedVideoConfigurations.
* No processed non-stalling or raw streams are configured.
When above conditions are NOT satistied, the controls of this mode and
android.control.aeTargetFpsRange will be ignored by the camera device,
the camera device will fall back to android.control.mode `==` AUTO,
and the returned capture result metadata will give the fps range choosen
by the camera device.
Switching into or out of this mode may trigger some camera ISP/sensor
reconfigurations, which may introduce extra latency. It is recommended that
the application avoids unnecessary scene mode switch as much as possible.
</notes>
</value>
<value optional="true" hidden="true">HDR
<notes>
Turn on custom high dynamic range (HDR) mode.
This is intended for LEGACY mode devices only;
HAL3+ camera devices should not implement this mode.
</notes>
</value>
</enum>
<description>
A camera mode optimized for conditions typical in a particular
capture setting.
</description>
<range>android.control.availableSceneModes</range>
<details>
This is the mode that that is active when
`android.control.mode == USE_SCENE_MODE`. Aside from FACE_PRIORITY,
these modes will disable android.control.aeMode,
android.control.awbMode, and android.control.afMode while in use.
The scene modes available for a given camera device are listed in
android.control.availableSceneModes.
The interpretation and implementation of these scene modes is left
to the implementor of the camera device. Their behavior will not be
consistent across all devices, and any given device may only implement
a subset of these modes.
</details>
<hal_details>
HAL implementations that include scene modes are expected to provide
the per-scene settings to use for android.control.aeMode,
android.control.awbMode, and android.control.afMode in
android.control.sceneModeOverrides.
For HIGH_SPEED_VIDEO mode, if it is included in android.control.availableSceneModes,
the HAL must list supported video size and fps range in
android.control.availableHighSpeedVideoConfigurations. For a given size, e.g.
1280x720, if the HAL has two different sensor configurations for normal streaming
mode and high speed streaming, when this scene mode is set/reset in a sequence of capture
requests, the HAL may have to switch between different sensor modes.
</hal_details>
<tag id="BC" />
</entry>
<entry name="videoStabilizationMode" type="byte" visibility="public"
enum="true" hwlevel="legacy">
<enum>
<value>OFF
<notes>
Video stabilization is disabled.
</notes></value>
<value>ON
<notes>
Video stabilization is enabled.
</notes></value>
</enum>
<description>Whether video stabilization is
active.</description>
<details>
Video stabilization automatically translates and scales images from the camera
in order to stabilize motion between consecutive frames.
If enabled, video stabilization can modify the
android.scaler.cropRegion to keep the video stream stabilized.
Switching between different video stabilization modes may take several frames
to initialize, the camera device will report the current mode in capture result
metadata. For example, When "ON" mode is requested, the video stabilization modes
in the first several capture results may still be "OFF", and it will become "ON"
when the initialization is done.
If a camera device supports both this mode and OIS (android.lens.opticalStabilizationMode),
turning both modes on may produce undesirable interaction, so it is recommended not to
enable both at the same time.
Some LEGACY camera devices will not support changing video stabilization
mode; in this case the value will always override to OFF.
</details>
<tag id="BC" />
</entry>
</controls>
<static>
<entry name="aeAvailableAntibandingModes" type="byte" visibility="public"
type_notes="list of enums" container="array" typedef="enumList"
hwlevel="legacy">
<array>
<size>n</size>
</array>
<description>
The set of auto-exposure antibanding modes that are
supported by this camera device.
</description>
<details>
Not all of the auto-exposure anti-banding modes may be
supported by a given camera device. This field lists the
valid anti-banding modes that the application may request
for this camera device; they must include AUTO.
</details>
<tag id="BC" />
</entry>
<entry name="aeAvailableModes" type="byte" visibility="public"
type_notes="list of enums" container="array" typedef="enumList"
hwlevel="legacy">
<array>
<size>n</size>
</array>
<description>
The set of auto-exposure modes that are supported by this
camera device.
</description>
<details>
Not all the auto-exposure modes may be supported by a
given camera device, especially if no flash unit is
available. This entry lists the valid modes for
android.control.aeMode for this camera device.
All camera devices support ON, and all camera devices with flash
units support ON_AUTO_FLASH and ON_ALWAYS_FLASH.
FULL mode camera devices always support OFF mode,
which enables application control of camera exposure time,
sensitivity, and frame duration.
LEGACY mode camera devices never support OFF mode.
LIMITED mode devices support OFF if they support the MANUAL_SENSOR
capability.
</details>
<tag id="BC" />
</entry>
<entry name="aeAvailableTargetFpsRanges" type="int32" visibility="public"
type_notes="list of pairs of frame rates"
container="array" typedef="rangeInt"
hwlevel="legacy">
<array>
<size>2</size>
<size>n</size>
</array>
<description>List of frame rate ranges supported by the
auto-exposure (AE) algorithm/hardware</description>
<tag id="BC" />
</entry>
<entry name="aeCompensationRange" type="int32" visibility="public"
container="array" typedef="rangeInt"
hwlevel="legacy">
<array>
<size>2</size>
</array>
<description>Maximum and minimum exposure compensation
setting, in counts of
android.control.aeCompensationStep.</description>
<range>At least (-2,2)/(exp compensation step
size)</range>
<tag id="BC" />
</entry>
<entry name="aeCompensationStep" type="rational" visibility="public"
hwlevel="legacy">
<description>Smallest step by which exposure compensation
can be changed</description>
<range>&lt;= 1/2</range>
<tag id="BC" />
</entry>
<entry name="afAvailableModes" type="byte" visibility="public"
type_notes="List of enums" container="array" typedef="enumList"
hwlevel="legacy">
<array>
<size>n</size>
</array>
<description>List of auto-focus (AF) modes that can be
selected with android.control.afMode.</description>
<details>
Not all the auto-focus modes may be supported by a
given camera device. This entry lists the valid modes for
android.control.afMode for this camera device.
All LIMITED and FULL mode camera devices will support OFF mode, and all
camera devices with adjustable focuser units
(`android.lens.info.minimumFocusDistance &gt; 0`) will support AUTO mode.
LEGACY devices will support OFF mode only if they support
focusing to infinity (by also setting android.lens.focusDistance to
`0.0f`).
</details>
<tag id="BC" />
</entry>
<entry name="availableEffects" type="byte" visibility="public"
type_notes="List of enums (android.control.effectMode)." container="array"
typedef="enumList" hwlevel="legacy">
<array>
<size>n</size>
</array>
<description>
List containing the subset of color effects
specified in android.control.effectMode that is supported by
this device.
</description>
<range>
Any subset of enums from those specified in
android.control.effectMode. OFF must be included in any subset.
</range>
<details>
This list contains the color effect modes that can be applied to
images produced by the camera device. Only modes that have
been fully implemented for the current device may be included here.
Implementations are not expected to be consistent across all devices.
If no color effect modes are available for a device, this should
simply be set to OFF.
A color effect will only be applied if
android.control.mode != OFF.
</details>
<tag id="BC" />
</entry>
<entry name="availableSceneModes" type="byte" visibility="public"
type_notes="List of enums (android.control.sceneMode)."
container="array" typedef="enumList" hwlevel="legacy">
<array>
<size>n</size>
</array>
<description>
List containing a subset of scene modes
specified in android.control.sceneMode.
</description>
<range>
Any subset of the enums specified in android.control.sceneMode
not including DISABLED, or solely DISABLED if no
scene modes are available. FACE_PRIORITY must be included
if face detection is supported (i.e.`android.statistics.info.maxFaceCount &gt; 0`).
</range>
<details>
This list contains scene modes that can be set for the camera device.
Only scene modes that have been fully implemented for the
camera device may be included here. Implementations are not expected
to be consistent across all devices. If no scene modes are supported
by the camera device, this will be set to `[DISABLED]`.
</details>
<tag id="BC" />
</entry>
<entry name="availableVideoStabilizationModes" type="byte"
visibility="public" type_notes="List of enums." container="array"
typedef="enumList" hwlevel="legacy">
<array>
<size>n</size>
</array>
<description>List of video stabilization modes that can
be supported</description>
<range>OFF must be included</range>
<tag id="BC" />
</entry>
<entry name="awbAvailableModes" type="byte" visibility="public"
type_notes="List of enums"
container="array" typedef="enumList" hwlevel="legacy">
<array>
<size>n</size>
</array>
<description>The set of auto-white-balance modes (android.control.awbMode)
that are supported by this camera device.</description>
<details>
Not all the auto-white-balance modes may be supported by a
given camera device. This entry lists the valid modes for
android.control.awbMode for this camera device.
All camera devices will support ON mode.
FULL mode camera devices will always support OFF mode,
which enables application control of white balance, by using
android.colorCorrection.transform and android.colorCorrection.gains
(android.colorCorrection.mode must be set to TRANSFORM_MATRIX).
</details>
<tag id="BC" />
</entry>
<entry name="maxRegions" type="int32" visibility="hidden"
container="array" hwlevel="legacy">
<array>
<size>3</size>
</array>
<description>
List of the maximum number of regions that can be used for metering in
auto-exposure (AE), auto-white balance (AWB), and auto-focus (AF);
this corresponds to the the maximum number of elements in
android.control.aeRegions, android.control.awbRegions,
and android.control.afRegions.
</description>
<range>
Value must be &amp;gt;= 0 for each element. For full-capability devices
this value must be &amp;gt;= 1 for AE and AF. The order of the elements is:
`(AE, AWB, AF)`.</range>
<tag id="BC" />
</entry>
<entry name="maxRegionsAe" type="int32" visibility="public"
synthetic="true" hwlevel="legacy">
<description>
List of the maximum number of regions that can be used for metering in
auto-exposure (AE);
this corresponds to the the maximum number of elements in
android.control.aeRegions.
</description>
<range>
Value will be &amp;gt;= 0. For FULL-capability devices, this
value will be &amp;gt;= 1.
</range>
<hal_details>This entry is private to the framework. Fill in
maxRegions to have this entry be automatically populated.
</hal_details>
</entry>
<entry name="maxRegionsAwb" type="int32" visibility="public"
synthetic="true" hwlevel="legacy">
<description>
List of the maximum number of regions that can be used for metering in
auto-white balance (AWB);
this corresponds to the the maximum number of elements in
android.control.awbRegions.
</description>
<range>
Value will be &amp;gt;= 0.
</range>
<hal_details>This entry is private to the framework. Fill in
maxRegions to have this entry be automatically populated.
</hal_details>
</entry>
<entry name="maxRegionsAf" type="int32" visibility="public"
synthetic="true" hwlevel="legacy">
<description>
List of the maximum number of regions that can be used for metering in
auto-focus (AF);
this corresponds to the the maximum number of elements in
android.control.afRegions.
</description>
<range>
Value will be &amp;gt;= 0. For FULL-capability devices, this
value will be &amp;gt;= 1.
</range>
<hal_details>This entry is private to the framework. Fill in
maxRegions to have this entry be automatically populated.
</hal_details>
</entry>
<entry name="sceneModeOverrides" type="byte" visibility="system"
container="array" hwlevel="limited">
<array>
<size>3</size>
<size>length(availableSceneModes)</size>
</array>
<description>
Ordered list of auto-exposure, auto-white balance, and auto-focus
settings to use with each available scene mode.
</description>
<range>
For each available scene mode, the list must contain three
entries containing the android.control.aeMode,
android.control.awbMode, and android.control.afMode values used
by the camera device. The entry order is `(aeMode, awbMode, afMode)`
where aeMode has the lowest index position.
</range>
<details>
When a scene mode is enabled, the camera device is expected
to override android.control.aeMode, android.control.awbMode,
and android.control.afMode with its preferred settings for
that scene mode.
The order of this list matches that of availableSceneModes,
with 3 entries for each mode. The overrides listed
for FACE_PRIORITY are ignored, since for that
mode the application-set android.control.aeMode,
android.control.awbMode, and android.control.afMode values are
used instead, matching the behavior when android.control.mode
is set to AUTO. It is recommended that the FACE_PRIORITY
overrides should be set to 0.
For example, if availableSceneModes contains
`(FACE_PRIORITY, ACTION, NIGHT)`, then the camera framework
expects sceneModeOverrides to have 9 entries formatted like:
`(0, 0, 0, ON_AUTO_FLASH, AUTO, CONTINUOUS_PICTURE,
ON_AUTO_FLASH, INCANDESCENT, AUTO)`.
</details>
<hal_details>
To maintain backward compatibility, this list will be made available
in the static metadata of the camera service. The camera service will
use these values to set android.control.aeMode,
android.control.awbMode, and android.control.afMode when using a scene
mode other than FACE_PRIORITY.
</hal_details>
<tag id="BC" />
</entry>
</static>
<dynamic>
<entry name="aePrecaptureId" type="int32" visibility="system" deprecated="true">
<description>The ID sent with the latest
CAMERA2_TRIGGER_PRECAPTURE_METERING call</description>
<details>Must be 0 if no
CAMERA2_TRIGGER_PRECAPTURE_METERING trigger received yet
by HAL. Always updated even if AE algorithm ignores the
trigger</details>
</entry>
<clone entry="android.control.aeAntibandingMode" kind="controls">
</clone>
<clone entry="android.control.aeExposureCompensation" kind="controls">
</clone>
<clone entry="android.control.aeLock" kind="controls">
</clone>
<clone entry="android.control.aeMode" kind="controls">
</clone>
<clone entry="android.control.aeRegions" kind="controls">
</clone>
<clone entry="android.control.aeTargetFpsRange" kind="controls">
</clone>
<clone entry="android.control.aePrecaptureTrigger" kind="controls">
</clone>
<entry name="aeState" type="byte" visibility="public" enum="true"
hwlevel="limited">
<enum>
<value>INACTIVE
<notes>AE is off or recently reset.
When a camera device is opened, it starts in
this state. This is a transient state, the camera device may skip reporting
this state in capture result.</notes></value>
<value>SEARCHING
<notes>AE doesn't yet have a good set of control values
for the current scene.
This is a transient state, the camera device may skip
reporting this state in capture result.</notes></value>
<value>CONVERGED
<notes>AE has a good set of control values for the
current scene.</notes></value>
<value>LOCKED
<notes>AE has been locked.</notes></value>
<value>FLASH_REQUIRED
<notes>AE has a good set of control values, but flash
needs to be fired for good quality still
capture.</notes></value>
<value>PRECAPTURE
<notes>AE has been asked to do a precapture sequence
and is currently executing it.
Precapture can be triggered through setting
android.control.aePrecaptureTrigger to START.
Once PRECAPTURE completes, AE will transition to CONVERGED
or FLASH_REQUIRED as appropriate. This is a transient
state, the camera device may skip reporting this state in
capture result.</notes></value>
</enum>
<description>Current state of the auto-exposure (AE) algorithm.</description>
<details>Switching between or enabling AE modes (android.control.aeMode) always
resets the AE state to INACTIVE. Similarly, switching between android.control.mode,
or android.control.sceneMode if `android.control.mode == USE_SCENE_MODE` resets all
the algorithm states to INACTIVE.
The camera device can do several state transitions between two results, if it is
allowed by the state transition table. For example: INACTIVE may never actually be
seen in a result.
The state in the result is the state for this image (in sync with this image): if
AE state becomes CONVERGED, then the image data associated with this result should
be good to use.
Below are state transition tables for different AE modes.
State | Transition Cause | New State | Notes
:------------:|:----------------:|:---------:|:-----------------------:
INACTIVE | | INACTIVE | Camera device auto exposure algorithm is disabled
When android.control.aeMode is AE_MODE_ON_*:
State | Transition Cause | New State | Notes
:-------------:|:--------------------------------------------:|:--------------:|:-----------------:
INACTIVE | Camera device initiates AE scan | SEARCHING | Values changing
INACTIVE | android.control.aeLock is ON | LOCKED | Values locked
SEARCHING | Camera device finishes AE scan | CONVERGED | Good values, not changing
SEARCHING | Camera device finishes AE scan | FLASH_REQUIRED | Converged but too dark w/o flash
SEARCHING | android.control.aeLock is ON | LOCKED | Values locked
CONVERGED | Camera device initiates AE scan | SEARCHING | Values changing
CONVERGED | android.control.aeLock is ON | LOCKED | Values locked
FLASH_REQUIRED | Camera device initiates AE scan | SEARCHING | Values changing
FLASH_REQUIRED | android.control.aeLock is ON | LOCKED | Values locked
LOCKED | android.control.aeLock is OFF | SEARCHING | Values not good after unlock
LOCKED | android.control.aeLock is OFF | CONVERGED | Values good after unlock
LOCKED | android.control.aeLock is OFF | FLASH_REQUIRED | Exposure good, but too dark
PRECAPTURE | Sequence done. android.control.aeLock is OFF | CONVERGED | Ready for high-quality capture
PRECAPTURE | Sequence done. android.control.aeLock is ON | LOCKED | Ready for high-quality capture
Any state | android.control.aePrecaptureTrigger is START | PRECAPTURE | Start AE precapture metering sequence
For the above table, the camera device may skip reporting any state changes that happen
without application intervention (i.e. mode switch, trigger, locking). Any state that
can be skipped in that manner is called a transient state.
For example, for above AE modes (AE_MODE_ON_*), in addition to the state transitions
listed in above table, it is also legal for the camera device to skip one or more
transient states between two results. See below table for examples:
State | Transition Cause | New State | Notes
:-------------:|:-----------------------------------------------------------:|:--------------:|:-----------------:
INACTIVE | Camera device finished AE scan | CONVERGED | Values are already good, transient states are skipped by camera device.
Any state | android.control.aePrecaptureTrigger is START, sequence done | FLASH_REQUIRED | Converged but too dark w/o flash after a precapture sequence, transient states are skipped by camera device.
Any state | android.control.aePrecaptureTrigger is START, sequence done | CONVERGED | Converged after a precapture sequence, transient states are skipped by camera device.
CONVERGED | Camera device finished AE scan | FLASH_REQUIRED | Converged but too dark w/o flash after a new scan, transient states are skipped by camera device.
FLASH_REQUIRED | Camera device finished AE scan | CONVERGED | Converged after a new scan, transient states are skipped by camera device.
</details>
</entry>
<clone entry="android.control.afMode" kind="controls">
</clone>
<clone entry="android.control.afRegions" kind="controls">
</clone>
<clone entry="android.control.afTrigger" kind="controls">
</clone>
<entry name="afState" type="byte" visibility="public" enum="true"
hwlevel="legacy">
<enum>
<value>INACTIVE
<notes>AF is off or has not yet tried to scan/been asked
to scan.
When a camera device is opened, it starts in this
state. This is a transient state, the camera device may
skip reporting this state in capture
result.</notes></value>
<value>PASSIVE_SCAN
<notes>AF is currently performing an AF scan initiated the
camera device in a continuous autofocus mode.
Only used by CONTINUOUS_* AF modes. This is a transient
state, the camera device may skip reporting this state in
capture result.</notes></value>
<value>PASSIVE_FOCUSED
<notes>AF currently believes it is in focus, but may
restart scanning at any time.
Only used by CONTINUOUS_* AF modes. This is a transient
state, the camera device may skip reporting this state in
capture result.</notes></value>
<value>ACTIVE_SCAN
<notes>AF is performing an AF scan because it was
triggered by AF trigger.
Only used by AUTO or MACRO AF modes. This is a transient
state, the camera device may skip reporting this state in
capture result.</notes></value>
<value>FOCUSED_LOCKED
<notes>AF believes it is focused correctly and has locked
focus.
This state is reached only after an explicit START AF trigger has been
sent (android.control.afTrigger), when good focus has been obtained.
The lens will remain stationary until the AF mode (android.control.afMode) is changed or
a new AF trigger is sent to the camera device (android.control.afTrigger).
</notes></value>
<value>NOT_FOCUSED_LOCKED
<notes>AF has failed to focus successfully and has locked
focus.
This state is reached only after an explicit START AF trigger has been
sent (android.control.afTrigger), when good focus cannot be obtained.
The lens will remain stationary until the AF mode (android.control.afMode) is changed or
a new AF trigger is sent to the camera device (android.control.afTrigger).
</notes></value>
<value>PASSIVE_UNFOCUSED
<notes>AF finished a passive scan without finding focus,
and may restart scanning at any time.
Only used by CONTINUOUS_* AF modes. This is a transient state, the camera
device may skip reporting this state in capture result.
LEGACY camera devices do not support this state. When a passive
scan has finished, it will always go to PASSIVE_FOCUSED.
</notes></value>
</enum>
<description>Current state of auto-focus (AF) algorithm.</description>
<details>
Switching between or enabling AF modes (android.control.afMode) always
resets the AF state to INACTIVE. Similarly, switching between android.control.mode,
or android.control.sceneMode if `android.control.mode == USE_SCENE_MODE` resets all
the algorithm states to INACTIVE.
The camera device can do several state transitions between two results, if it is
allowed by the state transition table. For example: INACTIVE may never actually be
seen in a result.
The state in the result is the state for this image (in sync with this image): if
AF state becomes FOCUSED, then the image data associated with this result should
be sharp.
Below are state transition tables for different AF modes.
When android.control.afMode is AF_MODE_OFF or AF_MODE_EDOF:
State | Transition Cause | New State | Notes
:------------:|:----------------:|:---------:|:-----------:
INACTIVE | | INACTIVE | Never changes
When android.control.afMode is AF_MODE_AUTO or AF_MODE_MACRO:
State | Transition Cause | New State | Notes
:-----------------:|:----------------:|:------------------:|:--------------:
INACTIVE | AF_TRIGGER | ACTIVE_SCAN | Start AF sweep, Lens now moving
ACTIVE_SCAN | AF sweep done | FOCUSED_LOCKED | Focused, Lens now locked
ACTIVE_SCAN | AF sweep done | NOT_FOCUSED_LOCKED | Not focused, Lens now locked
ACTIVE_SCAN | AF_CANCEL | INACTIVE | Cancel/reset AF, Lens now locked
FOCUSED_LOCKED | AF_CANCEL | INACTIVE | Cancel/reset AF
FOCUSED_LOCKED | AF_TRIGGER | ACTIVE_SCAN | Start new sweep, Lens now moving
NOT_FOCUSED_LOCKED | AF_CANCEL | INACTIVE | Cancel/reset AF
NOT_FOCUSED_LOCKED | AF_TRIGGER | ACTIVE_SCAN | Start new sweep, Lens now moving
Any state | Mode change | INACTIVE |
For the above table, the camera device may skip reporting any state changes that happen
without application intervention (i.e. mode switch, trigger, locking). Any state that
can be skipped in that manner is called a transient state.
For example, for these AF modes (AF_MODE_AUTO and AF_MODE_MACRO), in addition to the
state transitions listed in above table, it is also legal for the camera device to skip
one or more transient states between two results. See below table for examples:
State | Transition Cause | New State | Notes
:-----------------:|:----------------:|:------------------:|:--------------:
INACTIVE | AF_TRIGGER | FOCUSED_LOCKED | Focus is already good or good after a scan, lens is now locked.
INACTIVE | AF_TRIGGER | NOT_FOCUSED_LOCKED | Focus failed after a scan, lens is now locked.
FOCUSED_LOCKED | AF_TRIGGER | FOCUSED_LOCKED | Focus is already good or good after a scan, lens is now locked.
NOT_FOCUSED_LOCKED | AF_TRIGGER | FOCUSED_LOCKED | Focus is good after a scan, lens is not locked.
When android.control.afMode is AF_MODE_CONTINUOUS_VIDEO:
State | Transition Cause | New State | Notes
:-----------------:|:-----------------------------------:|:------------------:|:--------------:
INACTIVE | Camera device initiates new scan | PASSIVE_SCAN | Start AF scan, Lens now moving
INACTIVE | AF_TRIGGER | NOT_FOCUSED_LOCKED | AF state query, Lens now locked
PASSIVE_SCAN | Camera device completes current scan| PASSIVE_FOCUSED | End AF scan, Lens now locked
PASSIVE_SCAN | Camera device fails current scan | PASSIVE_UNFOCUSED | End AF scan, Lens now locked
PASSIVE_SCAN | AF_TRIGGER | FOCUSED_LOCKED | Immediate transition, if focus is good. Lens now locked
PASSIVE_SCAN | AF_TRIGGER | NOT_FOCUSED_LOCKED | Immediate transition, if focus is bad. Lens now locked
PASSIVE_SCAN | AF_CANCEL | INACTIVE | Reset lens position, Lens now locked
PASSIVE_FOCUSED | Camera device initiates new scan | PASSIVE_SCAN | Start AF scan, Lens now moving
PASSIVE_UNFOCUSED | Camera device initiates new scan | PASSIVE_SCAN | Start AF scan, Lens now moving
PASSIVE_FOCUSED | AF_TRIGGER | FOCUSED_LOCKED | Immediate transition, lens now locked
PASSIVE_UNFOCUSED | AF_TRIGGER | NOT_FOCUSED_LOCKED | Immediate transition, lens now locked
FOCUSED_LOCKED | AF_TRIGGER | FOCUSED_LOCKED | No effect
FOCUSED_LOCKED | AF_CANCEL | INACTIVE | Restart AF scan
NOT_FOCUSED_LOCKED | AF_TRIGGER | NOT_FOCUSED_LOCKED | No effect
NOT_FOCUSED_LOCKED | AF_CANCEL | INACTIVE | Restart AF scan
When android.control.afMode is AF_MODE_CONTINUOUS_PICTURE:
State | Transition Cause | New State | Notes
:-----------------:|:------------------------------------:|:------------------:|:--------------:
INACTIVE | Camera device initiates new scan | PASSIVE_SCAN | Start AF scan, Lens now moving
INACTIVE | AF_TRIGGER | NOT_FOCUSED_LOCKED | AF state query, Lens now locked
PASSIVE_SCAN | Camera device completes current scan | PASSIVE_FOCUSED | End AF scan, Lens now locked
PASSIVE_SCAN | Camera device fails current scan | PASSIVE_UNFOCUSED | End AF scan, Lens now locked
PASSIVE_SCAN | AF_TRIGGER | FOCUSED_LOCKED | Eventual transition once the focus is good. Lens now locked
PASSIVE_SCAN | AF_TRIGGER | NOT_FOCUSED_LOCKED | Eventual transition if cannot find focus. Lens now locked
PASSIVE_SCAN | AF_CANCEL | INACTIVE | Reset lens position, Lens now locked
PASSIVE_FOCUSED | Camera device initiates new scan | PASSIVE_SCAN | Start AF scan, Lens now moving
PASSIVE_UNFOCUSED | Camera device initiates new scan | PASSIVE_SCAN | Start AF scan, Lens now moving
PASSIVE_FOCUSED | AF_TRIGGER | FOCUSED_LOCKED | Immediate trans. Lens now locked
PASSIVE_UNFOCUSED | AF_TRIGGER | NOT_FOCUSED_LOCKED | Immediate trans. Lens now locked
FOCUSED_LOCKED | AF_TRIGGER | FOCUSED_LOCKED | No effect
FOCUSED_LOCKED | AF_CANCEL | INACTIVE | Restart AF scan
NOT_FOCUSED_LOCKED | AF_TRIGGER | NOT_FOCUSED_LOCKED | No effect
NOT_FOCUSED_LOCKED | AF_CANCEL | INACTIVE | Restart AF scan
When switch between AF_MODE_CONTINUOUS_* (CAF modes) and AF_MODE_AUTO/AF_MODE_MACRO
(AUTO modes), the initial INACTIVE or PASSIVE_SCAN states may be skipped by the
camera device. When a trigger is included in a mode switch request, the trigger
will be evaluated in the context of the new mode in the request.
See below table for examples:
State | Transition Cause | New State | Notes
:-----------:|:--------------------------------------:|:----------------------------------------:|:--------------:
any state | CAF-->AUTO mode switch | INACTIVE | Mode switch without trigger, initial state must be INACTIVE
any state | CAF-->AUTO mode switch with AF_TRIGGER | trigger-reachable states from INACTIVE | Mode switch with trigger, INACTIVE is skipped
any state | AUTO-->CAF mode switch | passively reachable states from INACTIVE | Mode switch without trigger, passive transient state is skipped
</details>
</entry>
<entry name="afTriggerId" type="int32" visibility="system" deprecated="true">
<description>The ID sent with the latest
CAMERA2_TRIGGER_AUTOFOCUS call</description>
<details>Must be 0 if no CAMERA2_TRIGGER_AUTOFOCUS trigger
received yet by HAL. Always updated even if AF algorithm
ignores the trigger</details>
</entry>
<clone entry="android.control.awbLock" kind="controls">
</clone>
<clone entry="android.control.awbMode" kind="controls">
</clone>
<clone entry="android.control.awbRegions" kind="controls">
</clone>
<clone entry="android.control.captureIntent" kind="controls">
</clone>
<entry name="awbState" type="byte" visibility="public" enum="true"
hwlevel="limited">
<enum>
<value>INACTIVE
<notes>AWB is not in auto mode, or has not yet started metering.
When a camera device is opened, it starts in this
state. This is a transient state, the camera device may
skip reporting this state in capture
result.</notes></value>
<value>SEARCHING
<notes>AWB doesn't yet have a good set of control
values for the current scene.
This is a transient state, the camera device
may skip reporting this state in capture result.</notes></value>
<value>CONVERGED
<notes>AWB has a good set of control values for the
current scene.</notes></value>
<value>LOCKED
<notes>AWB has been locked.
</notes></value>
</enum>
<description>Current state of auto-white balance (AWB) algorithm.</description>
<details>Switching between or enabling AWB modes (android.control.awbMode) always
resets the AWB state to INACTIVE. Similarly, switching between android.control.mode,
or android.control.sceneMode if `android.control.mode == USE_SCENE_MODE` resets all
the algorithm states to INACTIVE.
The camera device can do several state transitions between two results, if it is
allowed by the state transition table. So INACTIVE may never actually be seen in
a result.
The state in the result is the state for this image (in sync with this image): if
AWB state becomes CONVERGED, then the image data associated with this result should
be good to use.
Below are state transition tables for different AWB modes.
When `android.control.awbMode != AWB_MODE_AUTO`:
State | Transition Cause | New State | Notes
:------------:|:----------------:|:---------:|:-----------------------:
INACTIVE | |INACTIVE |Camera device auto white balance algorithm is disabled
When android.control.awbMode is AWB_MODE_AUTO:
State | Transition Cause | New State | Notes
:-------------:|:--------------------------------:|:-------------:|:-----------------:
INACTIVE | Camera device initiates AWB scan | SEARCHING | Values changing
INACTIVE | android.control.awbLock is ON | LOCKED | Values locked
SEARCHING | Camera device finishes AWB scan | CONVERGED | Good values, not changing
SEARCHING | android.control.awbLock is ON | LOCKED | Values locked
CONVERGED | Camera device initiates AWB scan | SEARCHING | Values changing
CONVERGED | android.control.awbLock is ON | LOCKED | Values locked
LOCKED | android.control.awbLock is OFF | SEARCHING | Values not good after unlock
For the above table, the camera device may skip reporting any state changes that happen
without application intervention (i.e. mode switch, trigger, locking). Any state that
can be skipped in that manner is called a transient state.
For example, for this AWB mode (AWB_MODE_AUTO), in addition to the state transitions
listed in above table, it is also legal for the camera device to skip one or more
transient states between two results. See below table for examples:
State | Transition Cause | New State | Notes
:-------------:|:--------------------------------:|:-------------:|:-----------------:
INACTIVE | Camera device finished AWB scan | CONVERGED | Values are already good, transient states are skipped by camera device.
LOCKED | android.control.awbLock is OFF | CONVERGED | Values good after unlock, transient states are skipped by camera device.
</details>
</entry>
<clone entry="android.control.effectMode" kind="controls">
</clone>
<clone entry="android.control.mode" kind="controls">
</clone>
<clone entry="android.control.sceneMode" kind="controls">
</clone>
<clone entry="android.control.videoStabilizationMode" kind="controls">
</clone>
</dynamic>
<static>
<entry name="availableHighSpeedVideoConfigurations" type="int32" visibility="hidden"
container="array" typedef="highSpeedVideoConfiguration" hwlevel="limited">
<array>
<size>4</size>
<size>n</size>
</array>
<description>
List of available high speed video size and fps range configurations
supported by the camera device, in the format of (width, height, fps_min, fps_max).
</description>
<range>
For each configuration, the fps_max &amp;gt;= 60fps.
</range>
<details>
When HIGH_SPEED_VIDEO is supported in android.control.availableSceneModes,
this metadata will list the supported high speed video size and fps range
configurations. All the sizes listed in this configuration will be a subset
of the sizes reported by StreamConfigurationMap#getOutputSizes for processed
non-stalling formats.
For the high speed video use case, where the application will set
android.control.sceneMode to HIGH_SPEED_VIDEO in capture requests, the application must
select the video size and fps range from this metadata to configure the recording and
preview streams and setup the recording requests. For example, if the application intends
to do high speed recording, it can select the maximum size reported by this metadata to
configure output streams. Once the size is selected, application can filter this metadata
by selected size and get the supported fps ranges, and use these fps ranges to setup the
recording requests. Note that for the use case of multiple output streams, application
must select one unique size from this metadata to use. Otherwise a request error might
occur.
For normal video recording use case, where some application will NOT set
android.control.sceneMode to HIGH_SPEED_VIDEO in capture requests, the fps ranges
reported in this metadata must not be used to setup capture requests, or it will cause
request error.
</details>
<hal_details>
All the sizes listed in this configuration will be a subset of the sizes reported by
android.scaler.availableStreamConfigurations for processed non-stalling output formats.
Note that for all high speed video configurations, HAL must be able to support a minimum
of two streams, though the application might choose to configure just one stream.
Since the HIGH_SPEED_VIDEO mode may be turned on for preview view only case, the preview
fps is bounded by device refresh rate (e.g. 60fps). For a given resolution, it is
recommended that this list includes some fps ranges (e.g. [30, 60]) that is suitable
for preview only streaming case.
</hal_details>
<tag id="V1" />
</entry>
</static>
</section>
<section name="demosaic">
<controls>
<entry name="mode" type="byte" enum="true">
<enum>
<value>FAST
<notes>Minimal or no slowdown of frame rate compared to
Bayer RAW output.</notes></value>
<value>HIGH_QUALITY
<notes>Improved processing quality but the frame rate is slowed down
relative to raw output.</notes></value>
</enum>
<description>Controls the quality of the demosaicing
processing.</description>
<tag id="FUTURE" />
</entry>
</controls>
</section>
<section name="edge">
<controls>
<entry name="mode" type="byte" visibility="public" enum="true" hwlevel="full">
<enum>
<value>OFF
<notes>No edge enhancement is applied.</notes></value>
<value>FAST
<notes>Apply edge enhancement at a quality level that does not slow down frame rate relative to sensor
output</notes></value>
<value>HIGH_QUALITY
<notes>Apply high-quality edge enhancement, at a cost of reducing output frame rate.
</notes></value>
</enum>
<description>Operation mode for edge
enhancement.</description>
<details>Edge/sharpness/detail enhancement. OFF means no
enhancement will be applied by the camera device.
This must be set to one of the modes listed in android.edge.availableEdgeModes.
FAST/HIGH_QUALITY both mean camera device determined enhancement
will be applied. HIGH_QUALITY mode indicates that the
camera device will use the highest-quality enhancement algorithms,
even if it slows down capture rate. FAST means the camera device will
not slow down capture rate when applying edge enhancement.</details>
<tag id="V1" />
</entry>
<entry name="strength" type="byte">
<description>Control the amount of edge enhancement
applied to the images</description>
<units>1-10; 10 is maximum sharpening</units>
<tag id="FUTURE" />
</entry>
</controls>
<static>
<entry name="availableEdgeModes" type="byte" visibility="public"
type_notes="list of enums" container="array" typedef="enumList"
hwlevel="full">
<array>
<size>n</size>
</array>
<description>
The set of edge enhancement modes supported by this camera device.
</description>
<details>
This tag lists the valid modes for android.edge.mode.
Full-capability camera devices must always support OFF and FAST.
</details>
<tag id="V1" />
</entry>
</static>
<dynamic>
<clone entry="android.edge.mode" kind="controls">
<tag id="V1" />
</clone>
</dynamic>
</section>
<section name="flash">
<controls>
<entry name="firingPower" type="byte">
<description>Power for flash firing/torch</description>
<units>10 is max power; 0 is no flash. Linear</units>
<range>0 - 10</range>
<details>Power for snapshot may use a different scale than
for torch mode. Only one entry for torch mode will be
used</details>
<tag id="FUTURE" />
</entry>
<entry name="firingTime" type="int64">
<description>Firing time of flash relative to start of
exposure</description>
<units>nanoseconds</units>
<range>0-(exposure time-flash duration)</range>
<details>Clamped to (0, exposure time - flash
duration).</details>
<tag id="FUTURE" />
</entry>
<entry name="mode" type="byte" visibility="public" enum="true" hwlevel="legacy">
<enum>
<value>OFF
<notes>
Do not fire the flash for this capture.
</notes>
</value>
<value>SINGLE
<notes>
If the flash is available and charged, fire flash
for this capture.
</notes>
</value>
<value>TORCH
<notes>
Transition flash to continuously on.
</notes>
</value>
</enum>
<description>The desired mode for for the camera device's flash control.</description>
<details>
This control is only effective when flash unit is available
(`android.flash.info.available == true`).
When this control is used, the android.control.aeMode must be set to ON or OFF.
Otherwise, the camera device auto-exposure related flash control (ON_AUTO_FLASH,
ON_ALWAYS_FLASH, or ON_AUTO_FLASH_REDEYE) will override this control.
When set to OFF, the camera device will not fire flash for this capture.
When set to SINGLE, the camera device will fire flash regardless of the camera
device's auto-exposure routine's result. When used in still capture case, this
control should be used along with auto-exposure (AE) precapture metering sequence
(android.control.aePrecaptureTrigger), otherwise, the image may be incorrectly exposed.
When set to TORCH, the flash will be on continuously. This mode can be used
for use cases such as preview, auto-focus assist, still capture, or video recording.
The flash status will be reported by android.flash.state in the capture result metadata.
</details>
<tag id="BC" />
</entry>
</controls>
<static>
<namespace name="info">
<entry name="available" type="byte" visibility="public" enum="true"
typedef="boolean" hwlevel="legacy">
<enum>
<value>FALSE</value>
<value>TRUE</value>
</enum>
<description>Whether this camera device has a
flash.</description>
<details>If no flash, none of the flash controls do
anything. All other metadata should return 0.</details>
<tag id="BC" />
</entry>
<entry name="chargeDuration" type="int64">
<description>Time taken before flash can fire
again</description>
<units>nanoseconds</units>
<range>0-1e9</range>
<details>1 second too long/too short for recharge? Should
this be power-dependent?</details>
<tag id="FUTURE" />
</entry>
</namespace>
<entry name="colorTemperature" type="byte">
<description>The x,y whitepoint of the
flash</description>
<units>pair of floats</units>
<range>0-1 for both</range>
<tag id="FUTURE" />
</entry>
<entry name="maxEnergy" type="byte">
<description>Max energy output of the flash for a full
power single flash</description>
<units>lumen-seconds</units>
<range>&amp;gt;= 0</range>
<tag id="FUTURE" />
</entry>
</static>
<dynamic>
<clone entry="android.flash.firingPower" kind="controls">
</clone>
<clone entry="android.flash.firingTime" kind="controls">
</clone>
<clone entry="android.flash.mode" kind="controls"></clone>
<entry name="state" type="byte" visibility="public" enum="true"
hwlevel="limited">
<enum>
<value>UNAVAILABLE
<notes>No flash on camera.</notes></value>
<value>CHARGING
<notes>Flash is charging and cannot be fired.</notes></value>
<value>READY
<notes>Flash is ready to fire.</notes></value>
<value>FIRED
<notes>Flash fired for this capture.</notes></value>
<value>PARTIAL
<notes>Flash partially illuminated this frame.
This is usually due to the next or previous frame having
the flash fire, and the flash spilling into this capture
due to hardware limitations.</notes></value>
</enum>
<description>Current state of the flash
unit.</description>
<details>
When the camera device doesn't have flash unit
(i.e. `android.flash.info.available == false`), this state will always be UNAVAILABLE.
Other states indicate the current flash status.
In certain conditions, this will be available on LEGACY devices:
* Flash-less cameras always return UNAVAILABLE.
* Using android.control.aeMode `==` ON_ALWAYS_FLASH
will always return FIRED.
* Using android.flash.mode `==` TORCH
will always return FIRED.
In all other conditions the state will not be available on
LEGACY devices (i.e. it will be `null`).
</details>
</entry>
</dynamic>
</section>
<section name="hotPixel">
<controls>
<entry name="mode" type="byte" visibility="public" enum="true">
<enum>
<value>OFF
<notes>
No hot pixel correction is applied.
The frame rate must not be reduced relative to sensor raw output
for this option.
The hotpixel map may be returned in android.statistics.hotPixelMap.
</notes>
</value>
<value>FAST
<notes>
Hot pixel correction is applied, without reducing frame
rate relative to sensor raw output.
The hotpixel map may be returned in android.statistics.hotPixelMap.
</notes>
</value>
<value>HIGH_QUALITY
<notes>
High-quality hot pixel correction is applied, at a cost
of reducing frame rate relative to sensor raw output.
The hotpixel map may be returned in android.statistics.hotPixelMap.
</notes>
</value>
</enum>
<description>
Set operational mode for hot pixel correction.
</description>
<details>
Valid modes for this camera device are listed in
android.hotPixel.availableHotPixelModes.
Hotpixel correction interpolates out, or otherwise removes, pixels
that do not accurately encode the incoming light (i.e. pixels that
are stuck at an arbitrary value).
</details>
<tag id="V1" />
<tag id="RAW" />
</entry>
</controls>
<static>
<entry name="availableHotPixelModes" type="byte" visibility="public"
type_notes="list of enums" container="array" typedef="enumList">
<array>
<size>n</size>
</array>
<description>
The set of hot pixel correction modes that are supported by this
camera device.
</description>
<details>
This tag lists valid modes for android.hotPixel.mode.
FULL mode camera devices will always support FAST.
</details>
<hal_details>
To avoid performance issues, there will be significantly fewer hot
pixels than actual pixels on the camera sensor.
</hal_details>
<tag id="V1" />
<tag id="RAW" />
</entry>
</static>
<dynamic>
<clone entry="android.hotPixel.mode" kind="controls">
<tag id="V1" />
<tag id="RAW" />
</clone>
</dynamic>
</section>
<section name="jpeg">
<controls>
<entry name="gpsLocation" type="byte" visibility="public" synthetic="true"
typedef="location" hwlevel="legacy">
<description>
A location object to use when generating image GPS metadata.
</description>
</entry>
<entry name="gpsCoordinates" type="double" visibility="hidden"
type_notes="latitude, longitude, altitude. First two in degrees, the third in meters"
container="array" hwlevel="legacy">
<array>
<size>3</size>
</array>
<description>GPS coordinates to include in output JPEG
EXIF</description>
<range>(-180 - 180], [-90,90], [-inf, inf]</range>
<tag id="BC" />
</entry>
<entry name="gpsProcessingMethod" type="byte" visibility="hidden"
typedef="string" hwlevel="legacy">
<description>32 characters describing GPS algorithm to
include in EXIF</description>
<units>UTF-8 null-terminated string</units>
<tag id="BC" />
</entry>
<entry name="gpsTimestamp" type="int64" visibility="hidden" hwlevel="legacy">
<description>Time GPS fix was made to include in
EXIF</description>
<units>UTC in seconds since January 1, 1970</units>
<tag id="BC" />
</entry>
<entry name="orientation" type="int32" visibility="public" hwlevel="legacy">
<description>Orientation of JPEG image to
write</description>
<units>Degrees in multiples of 90</units>
<range>0, 90, 180, 270</range>
<tag id="BC" />
</entry>
<entry name="quality" type="byte" visibility="public" hwlevel="legacy">
<description>Compression quality of the final JPEG
image.</description>
<range>1-100; larger is higher quality</range>
<details>85-95 is typical usage range.</details>
<tag id="BC" />
</entry>
<entry name="thumbnailQuality" type="byte" visibility="public" hwlevel="legacy">
<description>Compression quality of JPEG
thumbnail.</description>
<range>1-100; larger is higher quality</range>
<tag id="BC" />
</entry>
<entry name="thumbnailSize" type="int32" visibility="public"
container="array" typedef="size" hwlevel="legacy">
<array>
<size>2</size>
</array>
<description>Resolution of embedded JPEG thumbnail.</description>
<range>Size must be one of the size from android.jpeg.availableThumbnailSizes</range>
<details>When set to (0, 0) value, the JPEG EXIF will not contain thumbnail,
but the captured JPEG will still be a valid image.
When a jpeg image capture is issued, the thumbnail size selected should have
the same aspect ratio as the jpeg image.
If the thumbnail image aspect ratio differs from the JPEG primary image aspect
ratio, the camera device creates the thumbnail by cropping it from the primary image.
For example, if the primary image has 4:3 aspect ratio, the thumbnail image has
16:9 aspect ratio, the primary image will be cropped vertically (letterbox) to
generate the thumbnail image. The thumbnail image will always have a smaller Field
Of View (FOV) than the primary image when aspect ratios differ.
</details>
<hal_details>
The HAL must not squeeze or stretch the downscaled primary image to generate thumbnail.
The cropping must be done on the primary jpeg image rather than the sensor active array.
The stream cropping rule specified by "S5. Cropping" in camera3.h doesn't apply to the
thumbnail image cropping.
</hal_details>
<tag id="BC" />
</entry>
</controls>
<static>
<entry name="availableThumbnailSizes" type="int32" visibility="public"
container="array" typedef="size" hwlevel="legacy">
<array>
<size>2</size>
<size>n</size>
</array>
<description>Supported resolutions for the JPEG thumbnail.</description>
<range>Will include at least one valid resolution, plus
(0,0) for no thumbnail generation, and each size will be distinct.</range>
<details>Below condiditions will be satisfied for this size list:
* The sizes will be sorted by increasing pixel area (width x height).
If several resolutions have the same area, they will be sorted by increasing width.
* The aspect ratio of the largest thumbnail size will be same as the
aspect ratio of largest JPEG output size in android.scaler.availableStreamConfigurations.
The largest size is defined as the size that has the largest pixel area
in a given size list.
* Each output JPEG size in android.scaler.availableStreamConfigurations will have at least
one corresponding size that has the same aspect ratio in availableThumbnailSizes,
and vice versa.
* All non (0, 0) sizes will have non-zero widths and heights.</details>
<tag id="BC" />
</entry>
<entry name="maxSize" type="int32" visibility="system">
<description>Maximum size in bytes for the compressed
JPEG buffer</description>
<range>Must be large enough to fit any JPEG produced by
the camera</range>
<details>This is used for sizing the gralloc buffers for
JPEG</details>
</entry>
</static>
<dynamic>
<clone entry="android.jpeg.gpsLocation" kind="controls">
</clone>
<clone entry="android.jpeg.gpsCoordinates" kind="controls">
</clone>
<clone entry="android.jpeg.gpsProcessingMethod"
kind="controls"></clone>
<clone entry="android.jpeg.gpsTimestamp" kind="controls">
</clone>
<clone entry="android.jpeg.orientation" kind="controls">
</clone>
<clone entry="android.jpeg.quality" kind="controls">
</clone>
<entry name="size" type="int32">
<description>The size of the compressed JPEG image, in
bytes</description>
<range>&amp;gt;= 0</range>
<details>If no JPEG output is produced for the request,
this must be 0.
Otherwise, this describes the real size of the compressed
JPEG image placed in the output stream. More specifically,
if android.jpeg.maxSize = 1000000, and a specific capture
has android.jpeg.size = 500000, then the output buffer from
the JPEG stream will be 1000000 bytes, of which the first
500000 make up the real data.</details>
<tag id="FUTURE" />
</entry>
<clone entry="android.jpeg.thumbnailQuality"
kind="controls"></clone>
<clone entry="android.jpeg.thumbnailSize" kind="controls">
</clone>
</dynamic>
</section>
<section name="lens">
<controls>
<entry name="aperture" type="float" visibility="public" hwlevel="full">
<description>The ratio of lens focal length to the effective
aperture diameter.</description>
<units>f-number (f/NNN)</units>
<range>android.lens.info.availableApertures</range>
<details>This will only be supported on the camera devices that
have variable aperture lens. The aperture value can only be
one of the values listed in android.lens.info.availableApertures.
When this is supported and android.control.aeMode is OFF,
this can be set along with android.sensor.exposureTime,
android.sensor.sensitivity, and android.sensor.frameDuration
to achieve manual exposure control.
The requested aperture value may take several frames to reach the
requested value; the camera device will report the current (intermediate)
aperture size in capture result metadata while the aperture is changing.
While the aperture is still changing, android.lens.state will be set to MOVING.
When this is supported and android.control.aeMode is one of
the ON modes, this will be overridden by the camera device
auto-exposure algorithm, the overridden values are then provided
back to the user in the corresponding result.</details>
<tag id="V1" />
</entry>
<entry name="filterDensity" type="float" visibility="public" hwlevel="full">
<description>
State of lens neutral density filter(s).
</description>
<units>Steps of Exposure Value (EV).</units>
<range>android.lens.info.availableFilterDensities</range>
<details>
This will not be supported on most camera devices. On devices
where this is supported, this may only be set to one of the
values included in android.lens.info.availableFilterDensities.
Lens filters are typically used to lower the amount of light the
sensor is exposed to (measured in steps of EV). As used here, an EV
step is the standard logarithmic representation, which are
non-negative, and inversely proportional to the amount of light
hitting the sensor. For example, setting this to 0 would result
in no reduction of the incoming light, and setting this to 2 would
mean that the filter is set to reduce incoming light by two stops
(allowing 1/4 of the prior amount of light to the sensor).
It may take several frames before the lens filter density changes
to the requested value. While the filter density is still changing,
android.lens.state will be set to MOVING.
</details>
<tag id="V1" />
</entry>
<entry name="focalLength" type="float" visibility="public" hwlevel="legacy">
<description>
The current lens focal length; used for optical zoom.
</description>
<units>focal length in mm</units>
<range>android.lens.info.availableFocalLengths</range>
<details>
This setting controls the physical focal length of the camera
device's lens. Changing the focal length changes the field of
view of the camera device, and is usually used for optical zoom.
Like android.lens.focusDistance and android.lens.aperture, this
setting won't be applied instantaneously, and it may take several
frames before the lens can change to the requested focal length.
While the focal length is still changing, android.lens.state will
be set to MOVING.
This is expected not to be supported on most devices.
</details>
<tag id="V1" />
</entry>
<entry name="focusDistance" type="float" visibility="public" hwlevel="full">
<description>Distance to plane of sharpest focus,
measured from frontmost surface of the lens.</description>
<units>See android.lens.info.focusDistanceCalibration for details.</units>
<range>&amp;gt;= 0</range>
<details>`0.0f` means infinity focus. Used value will be clamped
to `[0.0f, android.lens.info.minimumFocusDistance]`.
Like android.lens.focalLength, this setting won't be applied
instantaneously, and it may take several frames before the lens
can move to the requested focus distance. While the lens is still moving,
android.lens.state will be set to MOVING.
LEGACY devices support at most setting this to `0.0f`
for infinity focus.
</details>
<tag id="BC" />
<tag id="V1" />
</entry>
<entry name="opticalStabilizationMode" type="byte" visibility="public"
enum="true" hwlevel="limited">
<enum>
<value>OFF
<notes>Optical stabilization is unavailable.</notes>
</value>
<value optional="true">ON
<notes>Optical stabilization is enabled.</notes>
</value>
</enum>
<description>
Sets whether the camera device uses optical image stabilization (OIS)
when capturing images.
</description>
<range>android.lens.info.availableOpticalStabilization</range>
<details>
OIS is used to compensate for motion blur due to small
movements of the camera during capture. Unlike digital image
stabilization (android.control.videoStabilizationMode), OIS
makes use of mechanical elements to stabilize the camera
sensor, and thus allows for longer exposure times before
camera shake becomes apparent.
Switching between different optical stabilization modes may take several
frames to initialize, the camera device will report the current mode in
capture result metadata. For example, When "ON" mode is requested, the
optical stabilization modes in the first several capture results may still
be "OFF", and it will become "ON" when the initialization is done.
If a camera device supports both OIS and EIS (android.control.videoStabilizationMode),
turning both modes on may produce undesirable interaction, so it is recommended not
to enable both at the same time.
Not all devices will support OIS; see
android.lens.info.availableOpticalStabilization for
available controls.
</details>
<tag id="V1" />
</entry>
</controls>
<static>
<namespace name="info">
<entry name="availableApertures" type="float" visibility="public"
container="array" hwlevel="full">
<array>
<size>n</size>
</array>
<description>List of supported aperture
values.</description>
<range>one entry required, &amp;gt; 0</range>
<details>If the camera device doesn't support variable apertures,
listed value will be the fixed aperture.
If the camera device supports variable apertures, the aperture value
in this list will be sorted in ascending order.</details>
<tag id="V1" />
</entry>
<entry name="availableFilterDensities" type="float" visibility="public"
container="array" hwlevel="full">
<array>
<size>n</size>
</array>
<description>
List of supported neutral density filter values for
android.lens.filterDensity.
</description>
<range>
At least one value is required. Values must be &amp;gt;= 0.
</range>
<details>
If changing android.lens.filterDensity is not supported,
availableFilterDensities must contain only 0. Otherwise, this
list contains only the exact filter density values available on
this camera device.
</details>
<tag id="V1" />
</entry>
<entry name="availableFocalLengths" type="float" visibility="public"
type_notes="The list of available focal lengths"
container="array" hwlevel="legacy">
<array>
<size>n</size>
</array>
<description>
The available focal lengths for this device for use with
android.lens.focalLength.
</description>
<range>
Each value in this list must be &amp;gt; 0. This list must
contain at least one value.
</range>
<details>
If optical zoom is not supported, this will only report
a single value corresponding to the static focal length of the
device. Otherwise, this will report every focal length supported
by the device.
</details>
<tag id="BC" />
<tag id="V1" />
</entry>
<entry name="availableOpticalStabilization" type="byte"
visibility="public" type_notes="list of enums" container="array"
typedef="enumList" hwlevel="limited">
<array>
<size>n</size>
</array>
<description>
List containing a subset of the optical image
stabilization (OIS) modes specified in
android.lens.opticalStabilizationMode.
</description>
<details>
If OIS is not implemented for a given camera device, this will
contain only OFF.
</details>
<tag id="V1" />
</entry>
<entry name="hyperfocalDistance" type="float" visibility="public" optional="true"
hwlevel="limited">
<description>Optional. Hyperfocal distance for this lens.</description>
<units>See android.lens.info.focusDistanceCalibration for details.</units>
<range>If lens is fixed focus, &amp;gt;= 0. If lens has focuser unit, the range is
`(0, android.lens.info.minimumFocusDistance]`</range>
<details>
If the lens is not fixed focus, the camera device will report this
field when android.lens.info.focusDistanceCalibration is APPROXIMATE or CALIBRATED.
</details>
</entry>
<entry name="minimumFocusDistance" type="float" visibility="public" optional="true"
hwlevel="limited">
<description>Shortest distance from frontmost surface
of the lens that can be focused correctly.</description>
<units>See android.lens.info.focusDistanceCalibration for details.</units>
<range>&amp;gt;= 0</range>
<details>If the lens is fixed-focus, this should be
0.</details>
<hal_details>Mandatory for FULL devices; LIMITED devices
must always set this value to 0 for fixed-focus; and may omit
the minimum focus distance otherwise.
This field is also mandatory for all devices advertising
the MANUAL_SENSOR capability.</hal_details>
<tag id="V1" />
</entry>
<entry name="shadingMapSize" type="int32" visibility="hidden"
type_notes="width and height (N, M) of lens shading map provided by the camera device."
container="array" typedef="size" hwlevel="full">
<array>
<size>2</size>
</array>
<description>Dimensions of lens shading map.</description>
<range>Both values &amp;gt;= 1</range>
<details>
The map should be on the order of 30-40 rows and columns, and
must be smaller than 64x64.
</details>
<tag id="V1" />
</entry>
<entry name="focusDistanceCalibration" type="byte" visibility="public"
enum="true" hwlevel="limited">
<enum>
<value>UNCALIBRATED
<notes>
The lens focus distance is not accurate, and the units used for
android.lens.focusDistance do not correspond to any physical units.
Setting the lens to the same focus distance on separate occasions may
result in a different real focus distance, depending on factors such
as the orientation of the device, the age of the focusing mechanism,
and the device temperature. The focus distance value will still be
in the range of `[0, android.lens.info.minimumFocusDistance]`, where 0
represents the farthest focus.
</notes>
</value>
<value>APPROXIMATE
<notes>
The lens focus distance is measured in diopters.
However, setting the lens to the same focus distance
on separate occasions may result in a different real
focus distance, depending on factors such as the
orientation of the device, the age of the focusing
mechanism, and the device temperature.
</notes>
</value>
<value>CALIBRATED
<notes>
The lens focus distance is measured in diopters, and
is calibrated.
The lens mechanism is calibrated so that setting the
same focus distance is repeatable on multiple
occasions with good accuracy, and the focus distance
corresponds to the real physical distance to the plane
of best focus.
</notes>
</value>
</enum>
<description>The lens focus distance calibration quality.</description>
<details>
The lens focus distance calibration quality determines the reliability of
focus related metadata entries, i.e. android.lens.focusDistance,
android.lens.focusRange, android.lens.info.hyperfocalDistance, and
android.lens.info.minimumFocusDistance.
</details>
<hal_details>
For devices advertise APPROXIMATE quality or higher, diopters 0 (infinity
focus) must work. When autofocus is disabled (android.control.afMode == OFF)
and the lens focus distance is set to 0 diopters
(android.lens.focusDistance == 0), the lens will move to focus at infinity
and is stably focused at infinity even if the device tilts. It may take the
lens some time to move; during the move the lens state should be MOVING and
the output diopter value should be changing toward 0.
</hal_details>
<tag id="V1" />
</entry>
</namespace>
<entry name="facing" type="byte" visibility="public" enum="true" hwlevel="legacy">
<enum>
<value>FRONT
<notes>
The camera device faces the same direction as the device's screen.
</notes></value>
<value>BACK
<notes>
The camera device faces the opposite direction as the device's screen.
</notes></value>
</enum>
<description>Direction the camera faces relative to
device screen.</description>
</entry>
<entry name="opticalAxisAngle" type="float"
type_notes="degrees. First defines the angle of separation between the perpendicular to the screen and the camera optical axis. The second then defines the clockwise rotation of the optical axis from native device up."
container="array">
<array>
<size>2</size>
</array>
<description>Relative angle of camera optical axis to the
perpendicular axis from the display</description>
<range>[0-90) for first angle, [0-360) for second</range>
<details>Examples:
(0,0) means that the camera optical axis
is perpendicular to the display surface;
(45,0) means that the camera points 45 degrees up when
device is held upright;
(45,90) means the camera points 45 degrees to the right when
the device is held upright.
Use FACING field to determine perpendicular outgoing
direction</details>
<tag id="FUTURE" />
</entry>
<entry name="position" type="float" container="array">
<array>
<size>3, location in mm, in the sensor coordinate
system</size>
</array>
<description>Coordinates of camera optical axis on
device</description>
<tag id="FUTURE" />
</entry>
</static>
<dynamic>
<clone entry="android.lens.aperture" kind="controls">
<tag id="V1" />
</clone>
<clone entry="android.lens.filterDensity" kind="controls">
<tag id="V1" />
</clone>
<clone entry="android.lens.focalLength" kind="controls">
<tag id="BC" />
</clone>
<clone entry="android.lens.focusDistance" kind="controls">
<details>Should be zero for fixed-focus cameras</details>
<tag id="BC" />
</clone>
<entry name="focusRange" type="float" visibility="public"
type_notes="Range of scene distances that are in focus"
container="array" typedef="pairFloatFloat" hwlevel="limited">
<array>
<size>2</size>
</array>
<description>The range of scene distances that are in
sharp focus (depth of field).</description>
<units>pair of focus distances in diopters: (near,
far), see android.lens.info.focusDistanceCalibration for details.</units>
<range>&amp;gt;=0</range>
<details>If variable focus not supported, can still report
fixed depth of field range</details>
<tag id="BC" />
</entry>
<clone entry="android.lens.opticalStabilizationMode"
kind="controls">
<tag id="V1" />
</clone>
<entry name="state" type="byte" visibility="public" enum="true" hwlevel="limited">
<enum>
<value>STATIONARY
<notes>
The lens parameters (android.lens.focalLength, android.lens.focusDistance,
android.lens.filterDensity and android.lens.aperture) are not changing.
</notes>
</value>
<value>MOVING
<notes>
One or several of the lens parameters
(android.lens.focalLength, android.lens.focusDistance,
android.lens.filterDensity or android.lens.aperture) is
currently changing.
</notes>
</value>
</enum>
<description>Current lens status.</description>
<details>
For lens parameters android.lens.focalLength, android.lens.focusDistance,
android.lens.filterDensity and android.lens.aperture, when changes are requested,
they may take several frames to reach the requested values. This state indicates
the current status of the lens parameters.
When the state is STATIONARY, the lens parameters are not changing. This could be
either because the parameters are all fixed, or because the lens has had enough
time to reach the most recently-requested values.
If all these lens parameters are not changable for a camera device, as listed below:
* Fixed focus (`android.lens.info.minimumFocusDistance == 0`), which means
android.lens.focusDistance parameter will always be 0.
* Fixed focal length (android.lens.info.availableFocalLengths contains single value),
which means the optical zoom is not supported.
* No ND filter (android.lens.info.availableFilterDensities contains only 0).
* Fixed aperture (android.lens.info.availableApertures contains single value).
Then this state will always be STATIONARY.
When the state is MOVING, it indicates that at least one of the lens parameters
is changing.
</details>
<tag id="V1" />
</entry>
</dynamic>
</section>
<section name="noiseReduction">
<controls>
<entry name="mode" type="byte" visibility="public" enum="true" hwlevel="full">
<enum>
<value>OFF
<notes>No noise reduction is applied.</notes></value>
<value>FAST
<notes>Noise reduction is applied without reducing frame rate relative to sensor
output.</notes></value>
<value>HIGH_QUALITY
<notes>High-quality noise reduction is applied, at the cost of reducing frame rate
relative to sensor output.</notes></value>
</enum>
<description>Mode of operation for the noise reduction algorithm.</description>
<details>Noise filtering control. OFF means no noise reduction
will be applied by the camera device.
This must be set to a valid mode from
android.noiseReduction.availableNoiseReductionModes.
FAST/HIGH_QUALITY both mean camera device determined noise filtering
will be applied. HIGH_QUALITY mode indicates that the camera device
will use the highest-quality noise filtering algorithms,
even if it slows down capture rate. FAST means the camera device will not
slow down capture rate when applying noise filtering.
</details>
<tag id="V1" />
</entry>
<entry name="strength" type="byte">
<description>Control the amount of noise reduction
applied to the images</description>
<units>1-10; 10 is max noise reduction</units>
<range>1 - 10</range>
<tag id="FUTURE" />
</entry>
</controls>
<static>
<entry name="availableNoiseReductionModes" type="byte" visibility="public"
type_notes="list of enums" container="array" typedef="enumList" hwlevel="limited">
<array>
<size>n</size>
</array>
<description>
The set of noise reduction modes supported by this camera device.
</description>
<details>
This tag lists the valid modes for android.noiseReduction.mode.
Full-capability camera devices must always support OFF and FAST.
Legacy-capability camera devices will only support FAST mode.
</details>
<tag id="V1" />
</entry>
</static>
<dynamic>
<clone entry="android.noiseReduction.mode" kind="controls">
<tag id="V1" />
</clone>
</dynamic>
</section>
<section name="quirks">
<static>
<entry name="meteringCropRegion" type="byte" visibility="system" deprecated="true" optional="true">
<description>If set to 1, the camera service does not
scale 'normalized' coordinates with respect to the crop
region. This applies to metering input (a{e,f,wb}Region
and output (face rectangles).</description>
<details>Normalized coordinates refer to those in the
(-1000,1000) range mentioned in the
android.hardware.Camera API.
HAL implementations should instead always use and emit
sensor array-relative coordinates for all region data. Does
not need to be listed in static metadata. Support will be
removed in future versions of camera service.</details>
</entry>
<entry name="triggerAfWithAuto" type="byte" visibility="system" deprecated="true" optional="true">
<description>If set to 1, then the camera service always
switches to FOCUS_MODE_AUTO before issuing a AF
trigger.</description>
<details>HAL implementations should implement AF trigger
modes for AUTO, MACRO, CONTINUOUS_FOCUS, and
CONTINUOUS_PICTURE modes instead of using this flag. Does
not need to be listed in static metadata. Support will be
removed in future versions of camera service</details>
</entry>
<entry name="useZslFormat" type="byte" visibility="system" deprecated="true" optional="true">
<description>If set to 1, the camera service uses
CAMERA2_PIXEL_FORMAT_ZSL instead of
HAL_PIXEL_FORMAT_IMPLEMENTATION_DEFINED for the zero
shutter lag stream</description>
<details>HAL implementations should use gralloc usage flags
to determine that a stream will be used for
zero-shutter-lag, instead of relying on an explicit
format setting. Does not need to be listed in static
metadata. Support will be removed in future versions of
camera service.</details>
</entry>
<entry name="usePartialResult" type="byte" visibility="hidden" deprecated="true" optional="true">
<description>
If set to 1, the HAL will always split result
metadata for a single capture into multiple buffers,
returned using multiple process_capture_result calls.
</description>
<details>
Does not need to be listed in static
metadata. Support for partial results will be reworked in
future versions of camera service. This quirk will stop
working at that point; DO NOT USE without careful
consideration of future support.
</details>
<hal_details>
Refer to `camera3_capture_result::partial_result`
for information on how to implement partial results.
</hal_details>
</entry>
</static>
<dynamic>
<entry name="partialResult" type="byte" visibility="hidden" deprecated="true" optional="true" enum="true" typedef="boolean">
<enum>
<value>FINAL
<notes>The last or only metadata result buffer
for this capture.</notes>
</value>
<value>PARTIAL
<notes>A partial buffer of result metadata for this
capture. More result buffers for this capture will be sent
by the camera device, the last of which will be marked
FINAL.</notes>
</value>
</enum>
<description>
Whether a result given to the framework is the
final one for the capture, or only a partial that contains a
subset of the full set of dynamic metadata
values.</description>
<range>Optional. Default value is FINAL.</range>
<details>
The entries in the result metadata buffers for a
single capture may not overlap, except for this entry. The
FINAL buffers must retain FIFO ordering relative to the
requests that generate them, so the FINAL buffer for frame 3 must
always be sent to the framework after the FINAL buffer for frame 2, and
before the FINAL buffer for frame 4. PARTIAL buffers may be returned
in any order relative to other frames, but all PARTIAL buffers for a given
capture must arrive before the FINAL buffer for that capture. This entry may
only be used by the camera device if quirks.usePartialResult is set to 1.
</details>
<hal_details>
Refer to `camera3_capture_result::partial_result`
for information on how to implement partial results.
</hal_details>
</entry>
</dynamic>
</section>
<section name="request">
<controls>
<entry name="frameCount" type="int32" visibility="system" deprecated="true">
<description>A frame counter set by the framework. Must
be maintained unchanged in output frame. This value monotonically
increases with every new result (that is, each new result has a unique
frameCount value).
</description>
<units>incrementing integer</units>
<range>Any int.</range>
</entry>
<entry name="id" type="int32" visibility="hidden">
<description>An application-specified ID for the current
request. Must be maintained unchanged in output
frame</description>
<units>arbitrary integer assigned by application</units>
<range>Any int</range>
<tag id="V1" />
</entry>
<entry name="inputStreams" type="int32" visibility="system" deprecated="true"
container="array">
<array>
<size>n</size>
</array>
<description>List which camera reprocess stream is used
for the source of reprocessing data.</description>
<units>List of camera reprocess stream IDs</units>
<range>
Typically, only one entry allowed, must be a valid reprocess stream ID.
</range>
<details>Only meaningful when android.request.type ==
REPROCESS. Ignored otherwise</details>
<tag id="HAL2" />
</entry>
<entry name="metadataMode" type="byte" visibility="system"
enum="true">
<enum>
<value>NONE
<notes>No metadata should be produced on output, except
for application-bound buffer data. If no
application-bound streams exist, no frame should be
placed in the output frame queue. If such streams
exist, a frame should be placed on the output queue
with null metadata but with the necessary output buffer
information. Timestamp information should still be
included with any output stream buffers</notes></value>
<value>FULL
<notes>All metadata should be produced. Statistics will
only be produced if they are separately
enabled</notes></value>
</enum>
<description>How much metadata to produce on
output</description>
<tag id="FUTURE" />
</entry>
<entry name="outputStreams" type="int32" visibility="system" deprecated="true"
container="array">
<array>
<size>n</size>
</array>
<description>Lists which camera output streams image data
from this capture must be sent to</description>
<units>List of camera stream IDs</units>
<range>List must only include streams that have been
created</range>
<details>If no output streams are listed, then the image
data should simply be discarded. The image data must
still be captured for metadata and statistics production,
and the lens and flash must operate as requested.</details>
<tag id="HAL2" />
</entry>
<entry name="type" type="byte" visibility="system" deprecated="true" enum="true">
<enum>
<value>CAPTURE
<notes>Capture a new image from the imaging hardware,
and process it according to the
settings</notes></value>
<value>REPROCESS
<notes>Process previously captured data; the
android.request.inputStreams parameter determines the
source reprocessing stream. TODO: Mark dynamic metadata
needed for reprocessing with [RP]</notes></value>
</enum>
<description>The type of the request; either CAPTURE or
REPROCESS. For HAL3, this tag is redundant.
</description>
<tag id="HAL2" />
</entry>
</controls>
<static>
<entry name="maxNumOutputStreams" type="int32" visibility="hidden"
container="array" hwlevel="legacy">
<array>
<size>3</size>
</array>
<description>The maximum numbers of different types of output streams
that can be configured and used simultaneously by a camera device.
</description>
<range>
For processed (and stalling) format streams, &amp;gt;= 1.
For Raw format (either stalling or non-stalling) streams, &amp;gt;= 0.
For processed (but not stalling) format streams, &amp;gt;= 3
for FULL mode devices (`android.info.supportedHardwareLevel == FULL`);
&amp;gt;= 2 for LIMITED mode devices (`android.info.supportedHardwareLevel == LIMITED`).
</range>
<details>
This is a 3 element tuple that contains the max number of output simultaneous
streams for raw sensor, processed (but not stalling), and processed (and stalling)
formats respectively. For example, assuming that JPEG is typically a processed and
stalling stream, if max raw sensor format output stream number is 1, max YUV streams
number is 3, and max JPEG stream number is 2, then this tuple should be `(1, 3, 2)`.
This lists the upper bound of the number of output streams supported by
the camera device. Using more streams simultaneously may require more hardware and
CPU resources that will consume more power. The image format for an output stream can
be any supported format provided by android.scaler.availableStreamConfigurations.
The formats defined in android.scaler.availableStreamConfigurations can be catergorized
into the 3 stream types as below:
* Processed (but stalling): any non-RAW format with a stallDurations &amp;gt; 0.
Typically JPEG format (ImageFormat#JPEG).
* Raw formats: ImageFormat#RAW_SENSOR, ImageFormat#RAW10 and ImageFormat#RAW_OPAQUE.
* Processed (but not-stalling): any non-RAW format without a stall duration.
Typically ImageFormat#YUV_420_888, ImageFormat#NV21, ImageFormat#YV12.
</details>
<tag id="BC" />
</entry>
<entry name="maxNumOutputRaw" type="int32" visibility="public" synthetic="true" hwlevel="legacy">
<description>The maximum numbers of different types of output streams
that can be configured and used simultaneously by a camera device
for any `RAW` formats.
</description>
<range>
&amp;gt;= 0
</range>
<details>
This value contains the max number of output simultaneous
streams from the raw sensor.
This lists the upper bound of the number of output streams supported by
the camera device. Using more streams simultaneously may require more hardware and
CPU resources that will consume more power. The image format for this kind of an output stream can
be any `RAW` and supported format provided by android.scaler.streamConfigurationMap.
In particular, a `RAW` format is typically one of:
* ImageFormat#RAW_SENSOR
* ImageFormat#RAW10
* Opaque `RAW`
LEGACY mode devices (android.info.supportedHardwareLevel `==` LEGACY)
never support raw streams.
</details>
</entry>
<entry name="maxNumOutputProc" type="int32" visibility="public" synthetic="true" hwlevel="legacy">
<description>The maximum numbers of different types of output streams
that can be configured and used simultaneously by a camera device
for any processed (but not-stalling) formats.
</description>
<range>
&amp;gt;= 3
for FULL mode devices (`android.info.supportedHardwareLevel == FULL`);
&amp;gt;= 2 for LIMITED mode devices (`android.info.supportedHardwareLevel == LIMITED`).
</range>
<details>
This value contains the max number of output simultaneous
streams for any processed (but not-stalling) formats.
This lists the upper bound of the number of output streams supported by
the camera device. Using more streams simultaneously may require more hardware and
CPU resources that will consume more power. The image format for this kind of an output stream can
be any non-`RAW` and supported format provided by android.scaler.streamConfigurationMap.
Processed (but not-stalling) is defined as any non-RAW format without a stall duration.
Typically:
* ImageFormat#YUV_420_888
* ImageFormat#NV21
* ImageFormat#YV12
* Implementation-defined formats, i.e. StreamConfiguration#isOutputSupportedFor(Class)
For full guarantees, query StreamConfigurationMap#getOutputStallDuration with
a processed format -- it will return 0 for a non-stalling stream.
LEGACY devices will support up to 3 processing/non-stalling streams.
</details>
</entry>
<entry name="maxNumOutputProcStalling" type="int32" visibility="public" synthetic="true" hwlevel="legacy">
<description>The maximum numbers of different types of output streams
that can be configured and used simultaneously by a camera device
for any processed (and stalling) formats.
</description>
<range>
&amp;gt;= 1
</range>
<details>
This value contains the max number of output simultaneous
streams for any processed (but not-stalling) formats.
This lists the upper bound of the number of output streams supported by
the camera device. Using more streams simultaneously may require more hardware and
CPU resources that will consume more power. The image format for this kind of an output stream can
be any non-`RAW` and supported format provided by android.scaler.streamConfigurationMap.
A processed and stalling format is defined as any non-RAW format with a stallDurations &amp;gt; 0.
Typically only the `JPEG` format (ImageFormat#JPEG)
For full guarantees, query StreamConfigurationMap#getOutputStallDuration with
a processed format -- it will return a non-0 value for a stalling stream.
LEGACY devices will support up to 1 processing/stalling stream.
</details>
</entry>
<entry name="maxNumReprocessStreams" type="int32" visibility="system"
deprecated="true" container="array">
<array>
<size>1</size>
</array>
<description>How many reprocessing streams of any type
can be allocated at the same time.</description>
<range>&amp;gt;= 0</range>
<details>
Only used by HAL2.x.
When set to 0, it means no reprocess stream is supported.
</details>
<tag id="HAL2" />
</entry>
<entry name="maxNumInputStreams" type="int32" visibility="hidden" hwlevel="full">
<description>
The maximum numbers of any type of input streams
that can be configured and used simultaneously by a camera device.
</description>
<range>
0 or 1.
</range>
<details>When set to 0, it means no input stream is supported.
The image format for a input stream can be any supported
format provided by
android.scaler.availableInputOutputFormatsMap. When using an
input stream, there must be at least one output stream
configured to to receive the reprocessed images.
For example, for Zero Shutter Lag (ZSL) still capture use case, the input
stream image format will be RAW_OPAQUE, the associated output stream image format
should be JPEG.
</details>
</entry>
</static>
<dynamic>
<entry name="frameCount" type="int32" visibility="hidden" deprecated="true">
<description>A frame counter set by the framework. This value monotonically
increases with every new result (that is, each new result has a unique
frameCount value).</description>
<units>count of frames</units>
<range>&amp;gt; 0</range>
<details>Reset on release()</details>
</entry>
<clone entry="android.request.id" kind="controls"></clone>
<clone entry="android.request.metadataMode"
kind="controls"></clone>
<clone entry="android.request.outputStreams"
kind="controls"></clone>
<entry name="pipelineDepth" type="byte" visibility="public" hwlevel="legacy">
<description>Specifies the number of pipeline stages the frame went
through from when it was exposed to when the final completed result
was available to the framework.</description>
<range>&amp;lt;= android.request.pipelineMaxDepth</range>
<details>Depending on what settings are used in the request, and
what streams are configured, the data may undergo less processing,
and some pipeline stages skipped.
See android.request.pipelineMaxDepth for more details.
</details>
<hal_details>
This value must always represent the accurate count of how many
pipeline stages were actually used.
</hal_details>
</entry>
</dynamic>
<static>
<entry name="pipelineMaxDepth" type="byte" visibility="public" hwlevel="legacy">
<description>Specifies the number of maximum pipeline stages a frame
has to go through from when it's exposed to when it's available
to the framework.</description>
<details>A typical minimum value for this is 2 (one stage to expose,
one stage to readout) from the sensor. The ISP then usually adds
its own stages to do custom HW processing. Further stages may be
added by SW processing.
Depending on what settings are used (e.g. YUV, JPEG) and what
processing is enabled (e.g. face detection), the actual pipeline
depth (specified by android.request.pipelineDepth) may be less than
the max pipeline depth.
A pipeline depth of X stages is equivalent to a pipeline latency of
X frame intervals.
This value will be 8 or less.
</details>
<hal_details>
This value should be 4 or less.
</hal_details>
</entry>
<entry name="partialResultCount" type="int32" visibility="public" optional="true">
<description>Defines how many sub-components
a result will be composed of.
</description>
<range>&amp;gt;= 1</range>
<details>In order to combat the pipeline latency, partial results
may be delivered to the application layer from the camera device as
soon as they are available.
Optional; defaults to 1. A value of 1 means that partial
results are not supported, and only the final TotalCaptureResult will
be produced by the camera device.
A typical use case for this might be: after requesting an
auto-focus (AF) lock the new AF state might be available 50%
of the way through the pipeline. The camera device could
then immediately dispatch this state via a partial result to
the application, and the rest of the metadata via later
partial results.
</details>
</entry>
<entry name="availableCapabilities" type="byte" visibility="public"
enum="true" container="array" hwlevel="legacy">
<array>
<size>n</size>
</array>
<enum>
<value>BACKWARD_COMPATIBLE
<notes>The minimal set of capabilities that every camera
device (regardless of android.info.supportedHardwareLevel)
supports.
This capability is listed by all devices, and
indicates that the camera device has a feature set
that's comparable to the baseline requirements for the
older android.hardware.Camera API.
</notes>
</value>
<value optional="true">MANUAL_SENSOR
<notes>
The camera device can be manually controlled (3A algorithms such
as auto-exposure, and auto-focus can be bypassed).
The camera device supports basic manual control of the sensor image
acquisition related stages. This means the following controls are
guaranteed to be supported:
* Manual frame duration control
* android.sensor.frameDuration
* android.sensor.info.maxFrameDuration
* Manual exposure control
* android.sensor.exposureTime
* android.sensor.info.exposureTimeRange
* Manual sensitivity control
* android.sensor.sensitivity
* android.sensor.info.sensitivityRange
* Manual lens control (if the lens is adjustable)
* android.lens.*
* Manual flash control (if a flash unit is present)
* android.flash.*
* Manual black level locking
* android.blackLevel.lock
If any of the above 3A algorithms are enabled, then the camera
device will accurately report the values applied by 3A in the
result.
A given camera device may also support additional manual sensor controls,
but this capability only covers the above list of controls.
If this is supported, android.scaler.streamConfigurationMap will
additionally return a min frame duration that is greater than
zero for each supported size-format combination.
</notes>
</value>
<value optional="true">MANUAL_POST_PROCESSING
<notes>
The camera device post-processing stages can be manually controlled.
The camera device supports basic manual control of the image post-processing
stages. This means the following controls are guaranteed to be supported:
* Manual tonemap control
* android.tonemap.curve
* android.tonemap.mode
* android.tonemap.maxCurvePoints
* Manual white balance control
* android.colorCorrection.transform
* android.colorCorrection.gains
* Manual lens shading map control
* android.shading.mode
* android.statistics.lensShadingMapMode
* android.statistics.lensShadingMap
* android.lens.info.shadingMapSize
* Manual aberration correction control (if aberration correction is supported)
* android.colorCorrection.aberrationMode
* android.colorCorrection.availableAberrationModes
If auto white balance is enabled, then the camera device
will accurately report the values applied by AWB in the result.
A given camera device may also support additional post-processing
controls, but this capability only covers the above list of controls.
</notes>
</value>
<value optional="true">RAW
<notes>
The camera device supports outputting RAW buffers and
metadata for interpreting them.
Devices supporting the RAW capability allow both for
saving DNG files, and for direct application processing of
raw sensor images.
* RAW_SENSOR is supported as an output format.
* The maximum available resolution for RAW_SENSOR streams
will match either the value in
android.sensor.info.pixelArraySize or
android.sensor.info.activeArraySize.
* All DNG-related optional metadata entries are provided
by the camera device.
</notes>
</value>
<value optional="true" hidden="true">ZSL
<notes>
The camera device supports the Zero Shutter Lag use case.
* At least one input stream can be used.
* RAW_OPAQUE is supported as an output/input format
* Using RAW_OPAQUE does not cause a frame rate drop
relative to the sensor's maximum capture rate (at that
resolution).
* RAW_OPAQUE will be reprocessable into both YUV_420_888
and JPEG formats.
* The maximum available resolution for RAW_OPAQUE streams
(both input/output) will match the maximum available
resolution of JPEG streams.
</notes>
</value>
</enum>
<description>List of capabilities that the camera device
advertises as fully supporting.</description>
<details>
A capability is a contract that the camera device makes in order
to be able to satisfy one or more use cases.
Listing a capability guarantees that the whole set of features
required to support a common use will all be available.
Using a subset of the functionality provided by an unsupported
capability may be possible on a specific camera device implementation;
to do this query each of android.request.availableRequestKeys,
android.request.availableResultKeys,
android.request.availableCharacteristicsKeys.
The following capabilities are guaranteed to be available on
android.info.supportedHardwareLevel `==` FULL devices:
* MANUAL_SENSOR
* MANUAL_POST_PROCESSING
Other capabilities may be available on either FULL or LIMITED
devices, but the application should query this field to be sure.
</details>
<hal_details>
Additional constraint details per-capability will be available
in the Compatibility Test Suite.
Minimum baseline requirements required for the
BACKWARD_COMPATIBLE capability are not explicitly listed.
Instead refer to "BC" tags and the camera CTS tests in the
android.hardware.camera2.cts package.
Listed controls that can be either request or result (e.g.
android.sensor.exposureTime) must be available both in the
request and the result in order to be considered to be
capability-compliant.
For example, if the HAL claims to support MANUAL control,
then exposure time must be configurable via the request _and_
the actual exposure applied must be available via
the result.
If MANUAL_SENSOR is omitted, the HAL may choose to omit the
android.scaler.availableMinFrameDurations static property entirely.
</hal_details>
</entry>
<entry name="availableRequestKeys" type="int32" visibility="hidden"
container="array" hwlevel="legacy">
<array>
<size>n</size>
</array>
<description>A list of all keys that the camera device has available
to use with CaptureRequest.</description>
<details>Attempting to set a key into a CaptureRequest that is not
listed here will result in an invalid request and will be rejected
by the camera device.
This field can be used to query the feature set of a camera device
at a more granular level than capabilities. This is especially
important for optional keys that are not listed under any capability
in android.request.availableCapabilities.
</details>
<hal_details>
Vendor tags must not be listed here. Use the vendor tag metadata
extensions C api instead (refer to camera3.h for more details).
Setting/getting vendor tags will be checked against the metadata
vendor extensions API and not against this field.
The HAL must not consume any request tags that are not listed either
here or in the vendor tag list.
The public camera2 API will always make the vendor tags visible
via CameraCharacteristics#getAvailableCaptureRequestKeys.
</hal_details>
</entry>
<entry name="availableResultKeys" type="int32" visibility="hidden"
container="array" hwlevel="legacy">
<array>
<size>n</size>
</array>
<description>A list of all keys that the camera device has available
to use with CaptureResult.</description>
<details>Attempting to get a key from a CaptureResult that is not
listed here will always return a `null` value. Getting a key from
a CaptureResult that is listed here must never return a `null`
value.
The following keys may return `null` unless they are enabled:
* android.statistics.lensShadingMap (non-null iff android.statistics.lensShadingMapMode == ON)
(Those sometimes-null keys should nevertheless be listed here
if they are available.)
This field can be used to query the feature set of a camera device
at a more granular level than capabilities. This is especially
important for optional keys that are not listed under any capability
in android.request.availableCapabilities.
</details>
<hal_details>
Tags listed here must always have an entry in the result metadata,
even if that size is 0 elements. Only array-type tags (e.g. lists,
matrices, strings) are allowed to have 0 elements.
Vendor tags must not be listed here. Use the vendor tag metadata
extensions C api instead (refer to camera3.h for more details).
Setting/getting vendor tags will be checked against the metadata
vendor extensions API and not against this field.
The HAL must not produce any result tags that are not listed either
here or in the vendor tag list.
The public camera2 API will always make the vendor tags visible
via CameraCharacteristics#getAvailableCaptureResultKeys.
</hal_details>
</entry>
<entry name="availableCharacteristicsKeys" type="int32" visibility="hidden"
container="array" hwlevel="legacy">
<array>
<size>n</size>
</array>
<description>A list of all keys that the camera device has available
to use with CameraCharacteristics.</description>
<details>This entry follows the same rules as
android.request.availableResultKeys (except that it applies for
CameraCharacteristics instead of CaptureResult). See above for more
details.
</details>
<hal_details>
Tags listed here must always have an entry in the static info metadata,
even if that size is 0 elements. Only array-type tags (e.g. lists,
matrices, strings) are allowed to have 0 elements.
Vendor tags must not be listed here. Use the vendor tag metadata
extensions C api instead (refer to camera3.h for more details).
Setting/getting vendor tags will be checked against the metadata
vendor extensions API and not against this field.
The HAL must not have any tags in its static info that are not listed
either here or in the vendor tag list.
The public camera2 API will always make the vendor tags visible
via CameraCharacteristics#getKeys.
</hal_details>
</entry>
</static>
</section>
<section name="scaler">
<controls>
<entry name="cropRegion" type="int32" visibility="public"
container="array" typedef="rectangle" hwlevel="legacy">
<array>
<size>4</size>
</array>
<description>The region of the sensor to read out for this capture.</description>
<units>(x,y) of top-left corner, width and height of region
in pixels; (0,0) is top-left corner of
android.sensor.info.activeArraySize</units>
<details>
The crop region coordinate system is based off
android.sensor.info.activeArraySize, with `(0, 0)` being the
top-left corner of the sensor active array.
Output streams use this rectangle to produce their output,
cropping to a smaller region if necessary to maintain the
stream's aspect ratio, then scaling the sensor input to
match the output's configured resolution.
The crop region is applied after the RAW to other color
space (e.g. YUV) conversion. Since raw streams
(e.g. RAW16) don't have the conversion stage, they are not
croppable. The crop region will be ignored by raw streams.
For non-raw streams, any additional per-stream cropping will
be done to maximize the final pixel area of the stream.
For example, if the crop region is set to a 4:3 aspect
ratio, then 4:3 streams will use the exact crop
region. 16:9 streams will further crop vertically
(letterbox).
Conversely, if the crop region is set to a 16:9, then 4:3
outputs will crop horizontally (pillarbox), and 16:9
streams will match exactly. These additional crops will
be centered within the crop region.
The width and height of the crop region cannot
be set to be smaller than
`floor( activeArraySize.width / android.scaler.availableMaxDigitalZoom )` and
`floor( activeArraySize.height / android.scaler.availableMaxDigitalZoom )`, respectively.
The camera device may adjust the crop region to account
for rounding and other hardware requirements; the final
crop region used will be included in the output capture
result.
</details>
<hal_details>
The output streams must maintain square pixels at all
times, no matter what the relative aspect ratios of the
crop region and the stream are. Negative values for
corner are allowed for raw output if full pixel array is
larger than active pixel array. Width and height may be
rounded to nearest larger supportable width, especially
for raw output, where only a few fixed scales may be
possible.
HAL2.x uses only (x, y, width)
</hal_details>
<tag id="BC" />
</entry>
</controls>
<static>
<entry name="availableFormats" type="int32"
visibility="hidden" deprecated="true" enum="true"
container="array" typedef="imageFormat">
<array>
<size>n</size>
</array>
<enum>
<value optional="true" id="0x20">RAW16
<notes>
RAW16 is a standard, cross-platform format for raw image
buffers with 16-bit pixels.
Buffers of this format are typically expected to have a
Bayer Color Filter Array (CFA) layout, which is given in
android.sensor.info.colorFilterArrangement. Sensors with
CFAs that are not representable by a format in
android.sensor.info.colorFilterArrangement should not
use this format.
Buffers of this format will also follow the constraints given for
RAW_OPAQUE buffers, but with relaxed performance constraints.
See android.scaler.availableInputOutputFormatsMap for
the full set of performance guarantees.
</notes>
</value>
<value optional="true" id="0x24">RAW_OPAQUE
<notes>
RAW_OPAQUE is a format for raw image buffers coming from an
image sensor.
The actual structure of buffers of this format is
platform-specific, but must follow several constraints:
1. No image post-processing operations may have been applied to
buffers of this type. These buffers contain raw image data coming
directly from the image sensor.
1. If a buffer of this format is passed to the camera device for
reprocessing, the resulting images will be identical to the images
produced if the buffer had come directly from the sensor and was
processed with the same settings.
The intended use for this format is to allow access to the native
raw format buffers coming directly from the camera sensor without
any additional conversions or decrease in framerate.
See android.scaler.availableInputOutputFormatsMap for the full set of
performance guarantees.
</notes>
</value>
<value optional="true" id="0x32315659">YV12
<notes>YCrCb 4:2:0 Planar</notes>
</value>
<value optional="true" id="0x11">YCrCb_420_SP
<notes>NV21</notes>
</value>
<value id="0x22">IMPLEMENTATION_DEFINED
<notes>System internal format, not application-accessible</notes>
</value>
<value id="0x23">YCbCr_420_888
<notes>Flexible YUV420 Format</notes>
</value>
<value id="0x21">BLOB
<notes>JPEG format</notes>
</value>
</enum>
<description>The list of image formats that are supported by this
camera device for output streams.</description>
<details>
All camera devices will support JPEG and YUV_420_888 formats.
When set to YUV_420_888, application can access the YUV420 data directly.
</details>
<hal_details>
These format values are from HAL_PIXEL_FORMAT_* in
system/core/include/system/graphics.h.
When IMPLEMENTATION_DEFINED is used, the platform
gralloc module will select a format based on the usage flags provided
by the camera HAL device and the other endpoint of the stream. It is
usually used by preview and recording streams, where the application doesn't
need access the image data.
YCbCr_420_888 format must be supported by the HAL. When an image stream
needs CPU/application direct access, this format will be used.
The BLOB format must be supported by the HAL. This is used for the JPEG stream.
A RAW_OPAQUE buffer should contain only pixel data. It is strongly
recommended that any information used by the camera device when
processing images is fully expressed by the result metadata
for that image buffer.
</hal_details>
<tag id="BC" />
</entry>
<entry name="availableJpegMinDurations" type="int64" visibility="hidden" deprecated="true"
container="array">
<array>
<size>n</size>
</array>
<description>The minimum frame duration that is supported
for each resolution in android.scaler.availableJpegSizes.
</description>
<units>ns</units>
<range>TODO: Remove property.</range>
<details>
This corresponds to the minimum steady-state frame duration when only
that JPEG stream is active and captured in a burst, with all
processing (typically in android.*.mode) set to FAST.
When multiple streams are configured, the minimum
frame duration will be &amp;gt;= max(individual stream min
durations)</details>
<tag id="BC" />
</entry>
<entry name="availableJpegSizes" type="int32" visibility="hidden"
deprecated="true" container="array" typedef="size">
<array>
<size>n</size>
<size>2</size>
</array>
<description>The JPEG resolutions that are supported by this camera device.</description>
<range>TODO: Remove property.</range>
<details>
The resolutions are listed as `(width, height)` pairs. All camera devices will support
sensor maximum resolution (defined by android.sensor.info.activeArraySize).
</details>
<hal_details>
The HAL must include sensor maximum resolution
(defined by android.sensor.info.activeArraySize),
and should include half/quarter of sensor maximum resolution.
</hal_details>
<tag id="BC" />
</entry>
<entry name="availableMaxDigitalZoom" type="float" visibility="public"
hwlevel="legacy">
<description>The maximum ratio between both active area width
and crop region width, and active area height and
crop region height.
This represents the maximum amount of zooming possible by
the camera device, or equivalently, the minimum cropping
window size.
Crop regions that have a width or height that is smaller
than this ratio allows will be rounded up to the minimum
allowed size by the camera device.
</description>
<range>&amp;gt;=1</range>
<tag id="BC" />
</entry>
<entry name="availableProcessedMinDurations" type="int64" visibility="hidden" deprecated="true"
container="array">
<array>
<size>n</size>
</array>
<description>For each available processed output size (defined in
android.scaler.availableProcessedSizes), this property lists the
minimum supportable frame duration for that size.
</description>
<units>ns</units>
<range>TODO: Remove property.</range>
<details>
This should correspond to the frame duration when only that processed
stream is active, with all processing (typically in android.*.mode)
set to FAST.
When multiple streams are configured, the minimum frame duration will
be &amp;gt;= max(individual stream min durations).
</details>
<tag id="BC" />
</entry>
<entry name="availableProcessedSizes" type="int32" visibility="hidden"
deprecated="true" container="array" typedef="size">
<array>
<size>n</size>
<size>2</size>
</array>
<description>The resolutions available for use with
processed output streams, such as YV12, NV12, and
platform opaque YUV/RGB streams to the GPU or video
encoders.</description>
<range>TODO: Remove property.</range>
<details>
The resolutions are listed as `(width, height)` pairs.
For a given use case, the actual maximum supported resolution
may be lower than what is listed here, depending on the destination
Surface for the image data. For example, for recording video,
the video encoder chosen may have a maximum size limit (e.g. 1080p)
smaller than what the camera (e.g. maximum resolution is 3264x2448)
can provide.
Please reference the documentation for the image data destination to
check if it limits the maximum size for image data.
</details>
<hal_details>
For FULL capability devices (`android.info.supportedHardwareLevel == FULL`),
the HAL must include all JPEG sizes listed in android.scaler.availableJpegSizes
and each below resolution if it is smaller than or equal to the sensor
maximum resolution (if they are not listed in JPEG sizes already):
* 240p (320 x 240)
* 480p (640 x 480)
* 720p (1280 x 720)
* 1080p (1920 x 1080)
For LIMITED capability devices (`android.info.supportedHardwareLevel == LIMITED`),
the HAL only has to list up to the maximum video size supported by the devices.
</hal_details>
<tag id="BC" />
</entry>
<entry name="availableRawMinDurations" type="int64" deprecated="true"
container="array">
<array>
<size>n</size>
</array>
<description>
For each available raw output size (defined in
android.scaler.availableRawSizes), this property lists the minimum
supportable frame duration for that size.
</description>
<units>ns</units>
<range>TODO: Remove property.</range>
<details>
Should correspond to the frame duration when only the raw stream is
active.
When multiple streams are configured, the minimum
frame duration will be &amp;gt;= max(individual stream min
durations)</details>
<tag id="BC" />
</entry>
<entry name="availableRawSizes" type="int32" deprecated="true"
container="array" typedef="size">
<array>
<size>n</size>
<size>2</size>
</array>
<description>The resolutions available for use with raw
sensor output streams, listed as width,
height</description>
<range>TODO: Remove property.
Must include: - sensor maximum resolution.</range>
</entry>
</static>
<dynamic>
<clone entry="android.scaler.cropRegion" kind="controls">
</clone>
</dynamic>
<static>
<entry name="availableInputOutputFormatsMap" type="int32"
visibility="hidden"
container="array" typedef="imageFormat"
hwlevel="full">
<array>
<size>n</size>
</array>
<description>The mapping of image formats that are supported by this
camera device for input streams, to their corresponding output formats.
</description>
<details>
All camera devices with at least 1
android.request.maxNumInputStreams will have at least one
available input format.
The camera device will support the following map of formats,
if its dependent capability is supported:
Input Format | Output Format | Capability
:---------------|:-----------------|:----------
RAW_OPAQUE | JPEG | ZSL
RAW_OPAQUE | YUV_420_888 | ZSL
RAW_OPAQUE | RAW16 | RAW
RAW16 | YUV_420_888 | RAW
RAW16 | JPEG | RAW
For ZSL-capable camera devices, using the RAW_OPAQUE format
as either input or output will never hurt maximum frame rate (i.e.
StreamConfigurationMap#getOutputStallDuration(int,Size)
for a `format =` RAW_OPAQUE is always 0).
Attempting to configure an input stream with output streams not
listed as available in this map is not valid.
TODO: typedef to ReprocessFormatMap
</details>
<hal_details>
For the formats, see `system/core/include/system/graphics.h` for a definition
of the image format enumerations.
This value is encoded as a variable-size array-of-arrays.
The inner array always contains `[format, length, ...]` where
`...` has `length` elements. An inner array is followed by another
inner array if the total metadata entry size hasn't yet been exceeded.
A code sample to read/write this encoding (with a device that
supports reprocessing RAW_OPAQUE to RAW16, YUV_420_888, and JPEG,
and reprocessing RAW16 to YUV_420_888 and JPEG):
// reading
int32_t* contents = &amp;entry.i32[0];
for (size_t i = 0; i &lt; entry.count; ) {
int32_t format = contents[i++];
int32_t length = contents[i++];
int32_t output_formats[length];
memcpy(&amp;output_formats[0], &amp;contents[i],
length * sizeof(int32_t));
i += length;
}
// writing (static example, RAW+ZSL)
int32_t[] contents = {
RAW_OPAQUE, 3, RAW16, YUV_420_888, BLOB,
RAW16, 2, YUV_420_888, BLOB,
};
update_camera_metadata_entry(metadata, index, &amp;contents[0],
sizeof(contents)/sizeof(contents[0]), &amp;updated_entry);
If the HAL claims to support any of the capabilities listed in the
above details, then it must also support all the input-output
combinations listed for that capability. It can optionally support
additional formats if it so chooses.
Refer to android.scaler.availableFormats for the enum values
which correspond to HAL_PIXEL_FORMAT_* in
system/core/include/system/graphics.h.
</hal_details>
</entry>
<entry name="availableStreamConfigurations" type="int32" visibility="hidden"
enum="true" container="array"
typedef="streamConfiguration" hwlevel="legacy">
<array>
<size>n</size>
<size>4</size>
</array>
<enum>
<value>OUTPUT</value>
<value>INPUT</value>
</enum>
<description>The available stream configurations that this
camera device supports
(i.e. format, width, height, output/input stream).
</description>
<details>
The configurations are listed as `(format, width, height, input?)`
tuples.
For a given use case, the actual maximum supported resolution
may be lower than what is listed here, depending on the destination
Surface for the image data. For example, for recording video,
the video encoder chosen may have a maximum size limit (e.g. 1080p)
smaller than what the camera (e.g. maximum resolution is 3264x2448)
can provide.
Please reference the documentation for the image data destination to
check if it limits the maximum size for image data.
Not all output formats may be supported in a configuration with
an input stream of a particular format. For more details, see
android.scaler.availableInputOutputFormatsMap.
The following table describes the minimum required output stream
configurations based on the hardware level
(android.info.supportedHardwareLevel):
Format | Size | Hardware Level | Notes
:-------------:|:--------------------------------------------:|:--------------:|:--------------:
JPEG | android.sensor.info.activeArraySize | Any |
JPEG | 1920x1080 (1080p) | Any | if 1080p &lt;= activeArraySize
JPEG | 1280x720 (720) | Any | if 720p &lt;= activeArraySize
JPEG | 640x480 (480p) | Any | if 480p &lt;= activeArraySize
JPEG | 320x240 (240p) | Any | if 240p &lt;= activeArraySize
YUV_420_888 | all output sizes available for JPEG | FULL |
YUV_420_888 | all output sizes available for JPEG, up to the maximum video size | LIMITED |
IMPLEMENTATION_DEFINED | same as YUV_420_888 | Any |
Refer to android.request.availableCapabilities for additional
mandatory stream configurations on a per-capability basis.
</details>
<hal_details>
It is recommended (but not mandatory) to also include half/quarter
of sensor maximum resolution for JPEG formats (regardless of hardware
level).
(The following is a rewording of the above required table):
For JPEG format, the sizes may be restricted by below conditions:
* The HAL may choose the aspect ratio of each Jpeg size to be one of well known ones
(e.g. 4:3, 16:9, 3:2 etc.). If the sensor maximum resolution
(defined by android.sensor.info.activeArraySize) has an aspect ratio other than these,
it does not have to be included in the supported JPEG sizes.
* Some hardware JPEG encoders may have pixel boundary alignment requirements, such as
the dimensions being a multiple of 16.
Therefore, the maximum JPEG size may be smaller than sensor maximum resolution.
However, the largest JPEG size must be as close as possible to the sensor maximum
resolution given above constraints. It is required that after aspect ratio adjustments,
additional size reduction due to other issues must be less than 3% in area. For example,
if the sensor maximum resolution is 3280x2464, if the maximum JPEG size has aspect
ratio 4:3, the JPEG encoder alignment requirement is 16, the maximum JPEG size will be
3264x2448.
For FULL capability devices (`android.info.supportedHardwareLevel == FULL`),
the HAL must include all YUV_420_888 sizes that have JPEG sizes listed
here as output streams.
It must also include each below resolution if it is smaller than or
equal to the sensor maximum resolution (for both YUV_420_888 and JPEG
formats), as output streams:
* 240p (320 x 240)
* 480p (640 x 480)
* 720p (1280 x 720)
* 1080p (1920 x 1080)
For LIMITED capability devices
(`android.info.supportedHardwareLevel == LIMITED`),
the HAL only has to list up to the maximum video size
supported by the device.
Regardless of hardware level, every output resolution available for
YUV_420_888 must also be available for IMPLEMENTATION_DEFINED.
This supercedes the following fields, which are now deprecated:
* availableFormats
* available[Processed,Raw,Jpeg]Sizes
</hal_details>
</entry>
<entry name="availableMinFrameDurations" type="int64" visibility="hidden"
container="array"
typedef="streamConfigurationDuration" hwlevel="legacy">
<array>
<size>4</size>
<size>n</size>
</array>
<description>This lists the minimum frame duration for each
format/size combination.
</description>
<units>(format, width, height, ns) x n</units>
<details>
This should correspond to the frame duration when only that
stream is active, with all processing (typically in android.*.mode)
set to either OFF or FAST.
When multiple streams are used in a request, the minimum frame
duration will be max(individual stream min durations).
The minimum frame duration of a stream (of a particular format, size)
is the same regardless of whether the stream is input or output.
See android.sensor.frameDuration and
android.scaler.availableStallDurations for more details about
calculating the max frame rate.
(Keep in sync with
StreamConfigurationMap#getOutputMinFrameDuration)
</details>
<tag id="V1" />
</entry>
<entry name="availableStallDurations" type="int64" visibility="hidden"
container="array" typedef="streamConfigurationDuration" hwlevel="legacy">
<array>
<size>4</size>
<size>n</size>
</array>
<description>This lists the maximum stall duration for each
format/size combination.
</description>
<units>(format, width, height, ns) x n</units>
<details>
A stall duration is how much extra time would get added
to the normal minimum frame duration for a repeating request
that has streams with non-zero stall.
For example, consider JPEG captures which have the following
characteristics:
* JPEG streams act like processed YUV streams in requests for which
they are not included; in requests in which they are directly
referenced, they act as JPEG streams. This is because supporting a
JPEG stream requires the underlying YUV data to always be ready for
use by a JPEG encoder, but the encoder will only be used (and impact
frame duration) on requests that actually reference a JPEG stream.
* The JPEG processor can run concurrently to the rest of the camera
pipeline, but cannot process more than 1 capture at a time.
In other words, using a repeating YUV request would result
in a steady frame rate (let's say it's 30 FPS). If a single
JPEG request is submitted periodically, the frame rate will stay
at 30 FPS (as long as we wait for the previous JPEG to return each
time). If we try to submit a repeating YUV + JPEG request, then
the frame rate will drop from 30 FPS.
In general, submitting a new request with a non-0 stall time
stream will _not_ cause a frame rate drop unless there are still
outstanding buffers for that stream from previous requests.
Submitting a repeating request with streams (call this `S`)
is the same as setting the minimum frame duration from
the normal minimum frame duration corresponding to `S`, added with
the maximum stall duration for `S`.
If interleaving requests with and without a stall duration,
a request will stall by the maximum of the remaining times
for each can-stall stream with outstanding buffers.
This means that a stalling request will not have an exposure start
until the stall has completed.
This should correspond to the stall duration when only that stream is
active, with all processing (typically in android.*.mode) set to FAST
or OFF. Setting any of the processing modes to HIGH_QUALITY
effectively results in an indeterminate stall duration for all
streams in a request (the regular stall calculation rules are
ignored).
The following formats may always have a stall duration:
* ImageFormat#JPEG
* ImageFormat#RAW_SENSOR
The following formats will never have a stall duration:
* ImageFormat#YUV_420_888
All other formats may or may not have an allowed stall duration on
a per-capability basis; refer to android.request.availableCapabilities
for more details.
See android.sensor.frameDuration for more information about
calculating the max frame rate (absent stalls).
(Keep up to date with
StreamConfigurationMap#getOutputStallDuration(int, Size) )
</details>
<hal_details>
If possible, it is recommended that all non-JPEG formats
(such as RAW16) should not have a stall duration. RAW10, RAW_OPAQUE and
IMPLEMENTATION_DEFINED must not have stall durations.
</hal_details>
<tag id="V1" />
</entry>
<entry name="streamConfigurationMap" type="int32" visibility="public"
synthetic="true" typedef="streamConfigurationMap"
hwlevel="legacy">
<description>The available stream configurations that this
camera device supports; also includes the minimum frame durations
and the stall durations for each format/size combination.
</description>
<details>
All camera devices will support sensor maximum resolution (defined by
android.sensor.info.activeArraySize) for the JPEG format.
For a given use case, the actual maximum supported resolution
may be lower than what is listed here, depending on the destination
Surface for the image data. For example, for recording video,
the video encoder chosen may have a maximum size limit (e.g. 1080p)
smaller than what the camera (e.g. maximum resolution is 3264x2448)
can provide.
Please reference the documentation for the image data destination to
check if it limits the maximum size for image data.
The following table describes the minimum required output stream
configurations based on the hardware level
(android.info.supportedHardwareLevel):
Format | Size | Hardware Level | Notes
:-------------:|:--------------------------------------------:|:--------------:|:--------------:
JPEG | android.sensor.info.activeArraySize | Any |
JPEG | 1920x1080 (1080p) | Any | if 1080p &lt;= activeArraySize
JPEG | 1280x720 (720) | Any | if 720p &lt;= activeArraySize
JPEG | 640x480 (480p) | Any | if 480p &lt;= activeArraySize
JPEG | 320x240 (240p) | Any | if 240p &lt;= activeArraySize
YUV_420_888 | all output sizes available for JPEG | FULL |
YUV_420_888 | all output sizes available for JPEG, up to the maximum video size | LIMITED |
IMPLEMENTATION_DEFINED | same as YUV_420_888 | Any |
Refer to android.request.availableCapabilities for additional
mandatory stream configurations on a per-capability basis.
</details>
<hal_details>
Do not set this property directly
(it is synthetic and will not be available at the HAL layer);
set the android.scaler.availableStreamConfigurations instead.
Not all output formats may be supported in a configuration with
an input stream of a particular format. For more details, see
android.scaler.availableInputOutputFormatsMap.
It is recommended (but not mandatory) to also include half/quarter
of sensor maximum resolution for JPEG formats (regardless of hardware
level).
(The following is a rewording of the above required table):
The HAL must include sensor maximum resolution (defined by
android.sensor.info.activeArraySize).
For FULL capability devices (`android.info.supportedHardwareLevel == FULL`),
the HAL must include all YUV_420_888 sizes that have JPEG sizes listed
here as output streams.
It must also include each below resolution if it is smaller than or
equal to the sensor maximum resolution (for both YUV_420_888 and JPEG
formats), as output streams:
* 240p (320 x 240)
* 480p (640 x 480)
* 720p (1280 x 720)
* 1080p (1920 x 1080)
For LIMITED capability devices
(`android.info.supportedHardwareLevel == LIMITED`),
the HAL only has to list up to the maximum video size
supported by the device.
Regardless of hardware level, every output resolution available for
YUV_420_888 must also be available for IMPLEMENTATION_DEFINED.
This supercedes the following fields, which are now deprecated:
* availableFormats
* available[Processed,Raw,Jpeg]Sizes
</hal_details>
</entry>
<entry name="croppingType" type="byte" visibility="public" enum="true"
hwlevel="legacy">
<enum>
<value>CENTER_ONLY
<notes>
The camera device only supports centered crop regions.
</notes>
</value>
<value>FREEFORM
<notes>
The camera device supports arbitrarily chosen crop regions.
</notes>
</value>
</enum>
<description>The crop type that this camera device supports.</description>
<details>
When passing a non-centered crop region (android.scaler.cropRegion) to a camera
device that only supports CENTER_ONLY cropping, the camera device will move the
crop region to the center of the sensor active array (android.sensor.info.activeArraySize)
and keep the crop region width and height unchanged. The camera device will return the
final used crop region in metadata result android.scaler.cropRegion.
Camera devices that support FREEFORM cropping will support any crop region that
is inside of the active array. The camera device will apply the same crop region and
return the final used crop region in capture result metadata android.scaler.cropRegion.
FULL capability devices (android.info.supportedHardwareLevel `==` FULL) will support
FREEFORM cropping. LEGACY capability devices will only support CENTER_ONLY cropping.
</details>
</entry>
</static>
</section>
<section name="sensor">
<controls>
<entry name="exposureTime" type="int64" visibility="public" hwlevel="full">
<description>Duration each pixel is exposed to
light.</description>
<units>nanoseconds</units>
<range>android.sensor.info.exposureTimeRange</range>
<details>If the sensor can't expose this exact duration, it should shorten the
duration exposed to the nearest possible value (rather than expose longer).
</details>
<tag id="V1" />
</entry>
<entry name="frameDuration" type="int64" visibility="public" hwlevel="full">
<description>Duration from start of frame exposure to
start of next frame exposure.</description>
<units>nanoseconds</units>
<range>See android.sensor.info.maxFrameDuration,
android.scaler.streamConfigurationMap. The duration
is capped to `max(duration, exposureTime + overhead)`.</range>
<details>
The maximum frame rate that can be supported by a camera subsystem is
a function of many factors:
* Requested resolutions of output image streams
* Availability of binning / skipping modes on the imager
* The bandwidth of the imager interface
* The bandwidth of the various ISP processing blocks
Since these factors can vary greatly between different ISPs and
sensors, the camera abstraction tries to represent the bandwidth
restrictions with as simple a model as possible.
The model presented has the following characteristics:
* The image sensor is always configured to output the smallest
resolution possible given the application's requested output stream
sizes. The smallest resolution is defined as being at least as large
as the largest requested output stream size; the camera pipeline must
never digitally upsample sensor data when the crop region covers the
whole sensor. In general, this means that if only small output stream
resolutions are configured, the sensor can provide a higher frame
rate.
* Since any request may use any or all the currently configured
output streams, the sensor and ISP must be configured to support
scaling a single capture to all the streams at the same time. This
means the camera pipeline must be ready to produce the largest
requested output size without any delay. Therefore, the overall
frame rate of a given configured stream set is governed only by the
largest requested stream resolution.
* Using more than one output stream in a request does not affect the
frame duration.
* Certain format-streams may need to do additional background processing
before data is consumed/produced by that stream. These processors
can run concurrently to the rest of the camera pipeline, but
cannot process more than 1 capture at a time.
The necessary information for the application, given the model above,
is provided via the android.scaler.streamConfigurationMap field
using StreamConfigurationMap#getOutputMinFrameDuration(int, Size).
These are used to determine the maximum frame rate / minimum frame
duration that is possible for a given stream configuration.
Specifically, the application can use the following rules to
determine the minimum frame duration it can request from the camera
device:
1. Let the set of currently configured input/output streams
be called `S`.
1. Find the minimum frame durations for each stream in `S`, by
looking it up in android.scaler.streamConfigurationMap using
StreamConfigurationMap#getOutputMinFrameDuration(int, Size) (with
its respective size/format). Let this set of frame durations be called
`F`.
1. For any given request `R`, the minimum frame duration allowed
for `R` is the maximum out of all values in `F`. Let the streams
used in `R` be called `S_r`.
If none of the streams in `S_r` have a stall time (listed in
StreamConfigurationMap#getOutputStallDuration(int,Size) using its
respective size/format), then the frame duration in
`F` determines the steady state frame rate that the application will
get if it uses `R` as a repeating request. Let this special kind
of request be called `Rsimple`.
A repeating request `Rsimple` can be _occasionally_ interleaved
by a single capture of a new request `Rstall` (which has at least
one in-use stream with a non-0 stall time) and if `Rstall` has the
same minimum frame duration this will not cause a frame rate loss
if all buffers from the previous `Rstall` have already been
delivered.
For more details about stalling, see
StreamConfigurationMap#getOutputStallDuration(int,Size).
</details>
<hal_details>
For more details about stalling, see
android.scaler.availableStallDurations.
</hal_details>
<tag id="V1" />
</entry>
<entry name="sensitivity" type="int32" visibility="public" hwlevel="full">
<description>The amount of gain applied to sensor data
before processing.</description>
<units>ISO arithmetic units</units>
<range>android.sensor.info.sensitivityRange</range>
<details>
The sensitivity is the standard ISO sensitivity value,
as defined in ISO 12232:2006.
The sensitivity must be within android.sensor.info.sensitivityRange, and
if if it less than android.sensor.maxAnalogSensitivity, the camera device
is guaranteed to use only analog amplification for applying the gain.
If the camera device cannot apply the exact sensitivity
requested, it will reduce the gain to the nearest supported
value. The final sensitivity used will be available in the
output capture result.
</details>
<hal_details>ISO 12232:2006 REI method is acceptable.</hal_details>
<tag id="V1" />
</entry>
</controls>
<static>
<namespace name="info">
<entry name="activeArraySize" type="int32" visibility="public"
type_notes="Four ints defining the active pixel rectangle"
container="array"
typedef="rectangle"
hwlevel="legacy">
<array>
<size>4</size>
</array>
<description>Area of raw data which corresponds to only
active pixels.</description>
<range>This array contains `(xmin, ymin, width, height)`. The `(xmin, ymin)` must be
&amp;gt;= `(0,0)`. The `(width, height)` must be &amp;lt;=
`android.sensor.info.pixelArraySize`.
</range>
<details>It is smaller or equal to
sensor full pixel array, which could include the black calibration pixels.</details>
<tag id="RAW" />
</entry>
<entry name="sensitivityRange" type="int32" visibility="public"
type_notes="Range of supported sensitivities"
container="array" typedef="rangeInt"
hwlevel="full">
<array>
<size>2</size>
</array>
<description>Range of valid sensitivities.</description>
<range>Min &lt;= 100, Max &amp;gt;= 800</range>
<details>
The minimum and maximum valid values for the
android.sensor.sensitivity control.
The values are the standard ISO sensitivity values,
as defined in ISO 12232:2006.
</details>
<tag id="BC" />
<tag id="V1" />
</entry>
<entry name="colorFilterArrangement" type="byte" visibility="public" enum="true"
hwlevel="full">
<enum>
<value>RGGB</value>
<value>GRBG</value>
<value>GBRG</value>
<value>BGGR</value>
<value>RGB
<notes>Sensor is not Bayer; output has 3 16-bit
values for each pixel, instead of just 1 16-bit value
per pixel.</notes></value>
</enum>
<description>The arrangement of color filters on sensor;
represents the colors in the top-left 2x2 section of
the sensor, in reading order.</description>
<tag id="RAW" />
</entry>
<entry name="exposureTimeRange" type="int64" visibility="public"
type_notes="nanoseconds" container="array" typedef="rangeLong"
hwlevel="full">
<array>
<size>2</size>
</array>
<description>Range of valid exposure
times used by android.sensor.exposureTime.</description>
<range>Min &lt;= 100e3 (100 us). For FULL capability devices
(android.info.supportedHardwareLevel == FULL), Max SHOULD be
&amp;gt;= 1e9 (1sec), MUST be &amp;gt;= 100e6 (100ms)</range>
<details>The min value will be &lt;= 100e3 (100 us). For FULL
capability devices (android.info.supportedHardwareLevel == FULL),
max will be &amp;gt;= 100e6 (100ms)</details>
<hal_details>For FULL capability devices (android.info.supportedHardwareLevel == FULL),
The maximum of the range SHOULD be at least
1 second (1e9), MUST be at least 100ms.</hal_details>
<tag id="V1" />
</entry>
<entry name="maxFrameDuration" type="int64" visibility="public"
hwlevel="full">
<description>Maximum possible frame duration (minimum frame
rate).</description>
<units>nanoseconds</units>
<range>For FULL capability devices
(android.info.supportedHardwareLevel == FULL), Max SHOULD be
&amp;gt;= 1e9 (1sec), MUST be &amp;gt;= 100e6 (100ms)
</range>
<details>The largest possible android.sensor.frameDuration
that will be accepted by the camera device. Attempting to use
frame durations beyond the maximum will result in the frame duration
being clipped to the maximum. See that control
for a full definition of frame durations.
Refer to
StreamConfigurationMap#getOutputMinFrameDuration(int,Size)
for the minimum frame duration values.
For FULL capability devices (android.info.supportedHardwareLevel == FULL),
max will be &amp;gt;= 100e6 (100ms).
</details>
<hal_details>
For FULL capability devices (android.info.supportedHardwareLevel == FULL),
The maximum of the range SHOULD be at least
1 second (1e9), MUST be at least 100ms (100e6).
android.sensor.info.maxFrameDuration must be greater or
equal to the android.sensor.info.exposureTimeRange max
value (since exposure time overrides frame duration).
Available minimum frame durations for JPEG must be no greater
than that of the YUV_420_888/IMPLEMENTATION_DEFINED
minimum frame durations (for that respective size).
Since JPEG processing is considered offline and can take longer than
a single uncompressed capture, refer to
android.scaler.availableStallDurations
for details about encoding this scenario.
</hal_details>
<tag id="V1" />
</entry>
<entry name="physicalSize" type="float" visibility="public"
type_notes="width x height in millimeters"
container="array" typedef="sizeF" hwlevel="legacy">
<array>
<size>2</size>
</array>
<description>The physical dimensions of the full pixel
array.</description>
<details>This is the physical size of the sensor pixel
array defined by android.sensor.info.pixelArraySize.
</details>
<hal_details>Needed for FOV calculation for old API</hal_details>
<tag id="V1" />
<tag id="BC" />
</entry>
<entry name="pixelArraySize" type="int32" visibility="public"
container="array" typedef="size" hwlevel="legacy">
<array>
<size>2</size>
</array>
<description>Dimensions of full pixel array, possibly
including black calibration pixels.</description>
<details>The pixel count of the full pixel array,
which covers android.sensor.info.physicalSize area.
If a camera device supports raw sensor formats, either this
or android.sensor.info.activeArraySize is the maximum output
raw size listed in android.scaler.streamConfigurationMap.
If a size corresponding to pixelArraySize is listed, the resulting
raw sensor image will include black pixels.
</details>
<tag id="RAW" />
<tag id="BC" />
</entry>
<entry name="whiteLevel" type="int32" visibility="public">
<description>
Maximum raw value output by sensor.
</description>
<range>&amp;gt; 255 (8-bit output)</range>
<details>
This specifies the fully-saturated encoding level for the raw
sample values from the sensor. This is typically caused by the
sensor becoming highly non-linear or clipping. The minimum for
each channel is specified by the offset in the
android.sensor.blackLevelPattern tag.
The white level is typically determined either by sensor bit depth
(8-14 bits is expected), or by the point where the sensor response
becomes too non-linear to be useful. The default value for this is
maximum representable value for a 16-bit raw sample (2^16 - 1).
</details>
<hal_details>
The full bit depth of the sensor must be available in the raw data,
so the value for linear sensors should not be significantly lower
than maximum raw value supported, i.e. 2^(sensor bits per pixel).
</hal_details>
<tag id="RAW" />
</entry>
<entry name="timestampSource" type="byte" visibility="public"
enum="true" hwlevel="legacy">
<enum>
<value>UNKNOWN
<notes>
Timestamps from android.sensor.timestamp are in nanoseconds and monotonic,
but can not be compared to timestamps from other subsystems
(e.g. accelerometer, gyro etc.), or other instances of the same or different
camera devices in the same system. Timestamps between streams and results for
a single camera instance are comparable, and the timestamps for all buffers
and the result metadata generated by a single capture are identical.
</notes>
</value>
<value>REALTIME
<notes>
Timestamps from android.sensor.timestamp are in the same timebase as
android.os.SystemClock#elapsedRealtimeNanos(),
and they can be compared to other timestamps using that base.
</notes>
</value>
</enum>
<description>The time base source for sensor capture start timestamps.</description>
<details>
The timestamps provided for captures are always in nanoseconds and monotonic, but
may not based on a time source that can be compared to other system time sources.
This characteristic defines the source for the timestamps, and therefore whether they
can be compared against other system time sources/timestamps.
</details>
<tag id="V1" />
</entry>
</namespace>
<entry name="referenceIlluminant1" type="byte" visibility="public"
enum="true">
<enum>
<value id="1">DAYLIGHT</value>
<value id="2">FLUORESCENT</value>
<value id="3">TUNGSTEN
<notes>Incandescent light</notes>
</value>
<value id="4">FLASH</value>
<value id="9">FINE_WEATHER</value>
<value id="10">CLOUDY_WEATHER</value>
<value id="11">SHADE</value>
<value id="12">DAYLIGHT_FLUORESCENT
<notes>D 5700 - 7100K</notes>
</value>
<value id="13">DAY_WHITE_FLUORESCENT
<notes>N 4600 - 5400K</notes>
</value>
<value id="14">COOL_WHITE_FLUORESCENT
<notes>W 3900 - 4500K</notes>
</value>
<value id="15">WHITE_FLUORESCENT
<notes>WW 3200 - 3700K</notes>
</value>
<value id="17">STANDARD_A</value>
<value id="18">STANDARD_B</value>
<value id="19">STANDARD_C</value>
<value id="20">D55</value>
<value id="21">D65</value>
<value id="22">D75</value>
<value id="23">D50</value>
<value id="24">ISO_STUDIO_TUNGSTEN</value>
</enum>
<description>
The standard reference illuminant used as the scene light source when
calculating the android.sensor.colorTransform1,
android.sensor.calibrationTransform1, and
android.sensor.forwardMatrix1 matrices.
</description>
<details>
The values in this tag correspond to the values defined for the
EXIF LightSource tag. These illuminants are standard light sources
that are often used calibrating camera devices.
If this tag is present, then android.sensor.colorTransform1,
android.sensor.calibrationTransform1, and
android.sensor.forwardMatrix1 will also be present.
Some devices may choose to provide a second set of calibration
information for improved quality, including
android.sensor.referenceIlluminant2 and its corresponding matrices.
</details>
<hal_details>
The first reference illuminant (android.sensor.referenceIlluminant1)
and corresponding matrices must be present to support the RAW capability
and DNG output.
When producing raw images with a color profile that has only been
calibrated against a single light source, it is valid to omit
android.sensor.referenceIlluminant2 along with the
android.sensor.colorTransform2, android.sensor.calibrationTransform2,
and android.sensor.forwardMatrix2 matrices.
If only android.sensor.referenceIlluminant1 is included, it should be
chosen so that it is representative of typical scene lighting. In
general, D50 or DAYLIGHT will be chosen for this case.
If both android.sensor.referenceIlluminant1 and
android.sensor.referenceIlluminant2 are included, they should be
chosen to represent the typical range of scene lighting conditions.
In general, low color temperature illuminant such as Standard-A will
be chosen for the first reference illuminant and a higher color
temperature illuminant such as D65 will be chosen for the second
reference illuminant.
</hal_details>
<tag id="RAW" />
</entry>
<entry name="referenceIlluminant2" type="byte" visibility="public">
<description>
The standard reference illuminant used as the scene light source when
calculating the android.sensor.colorTransform2,
android.sensor.calibrationTransform2, and
android.sensor.forwardMatrix2 matrices.
</description>
<details>
See android.sensor.referenceIlluminant1 for more details.
Valid values for this are the same as those given for the first
reference illuminant.
If this tag is present, then android.sensor.colorTransform2,
android.sensor.calibrationTransform2, and
android.sensor.forwardMatrix2 will also be present.
</details>
<tag id="RAW" />
</entry>
<entry name="calibrationTransform1" type="rational"
visibility="public" optional="true"
type_notes="3x3 matrix in row-major-order" container="array"
typedef="colorSpaceTransform">
<array>
<size>3</size>
<size>3</size>
</array>
<description>
A per-device calibration transform matrix that maps from the
reference sensor colorspace to the actual device sensor colorspace.
</description>
<details>
This matrix is used to correct for per-device variations in the
sensor colorspace, and is used for processing raw buffer data.
The matrix is expressed as a 3x3 matrix in row-major-order, and
contains a per-device calibration transform that maps colors
from reference sensor color space (i.e. the "golden module"
colorspace) into this camera device's native sensor color
space under the first reference illuminant
(android.sensor.referenceIlluminant1).
</details>
<tag id="RAW" />
</entry>
<entry name="calibrationTransform2" type="rational"
visibility="public" optional="true"
type_notes="3x3 matrix in row-major-order" container="array"
typedef="colorSpaceTransform">
<array>
<size>3</size>
<size>3</size>
</array>
<description>
A per-device calibration transform matrix that maps from the
reference sensor colorspace to the actual device sensor colorspace
(this is the colorspace of the raw buffer data).
</description>
<details>
This matrix is used to correct for per-device variations in the
sensor colorspace, and is used for processing raw buffer data.
The matrix is expressed as a 3x3 matrix in row-major-order, and
contains a per-device calibration transform that maps colors
from reference sensor color space (i.e. the "golden module"
colorspace) into this camera device's native sensor color
space under the second reference illuminant
(android.sensor.referenceIlluminant2).
This matrix will only be present if the second reference
illuminant is present.
</details>
<tag id="RAW" />
</entry>
<entry name="colorTransform1" type="rational"
visibility="public" optional="true"
type_notes="3x3 matrix in row-major-order" container="array"
typedef="colorSpaceTransform">
<array>
<size>3</size>
<size>3</size>
</array>
<description>
A matrix that transforms color values from CIE XYZ color space to
reference sensor color space.
</description>
<details>
This matrix is used to convert from the standard CIE XYZ color
space to the reference sensor colorspace, and is used when processing
raw buffer data.
The matrix is expressed as a 3x3 matrix in row-major-order, and
contains a color transform matrix that maps colors from the CIE
XYZ color space to the reference sensor color space (i.e. the
"golden module" colorspace) under the first reference illuminant
(android.sensor.referenceIlluminant1).
The white points chosen in both the reference sensor color space
and the CIE XYZ colorspace when calculating this transform will
match the standard white point for the first reference illuminant
(i.e. no chromatic adaptation will be applied by this transform).
</details>
<tag id="RAW" />
</entry>
<entry name="colorTransform2" type="rational"
visibility="public" optional="true"
type_notes="3x3 matrix in row-major-order" container="array"
typedef="colorSpaceTransform">
<array>
<size>3</size>
<size>3</size>
</array>
<description>
A matrix that transforms color values from CIE XYZ color space to
reference sensor color space.
</description>
<details>
This matrix is used to convert from the standard CIE XYZ color
space to the reference sensor colorspace, and is used when processing
raw buffer data.
The matrix is expressed as a 3x3 matrix in row-major-order, and
contains a color transform matrix that maps colors from the CIE
XYZ color space to the reference sensor color space (i.e. the
"golden module" colorspace) under the second reference illuminant
(android.sensor.referenceIlluminant2).
The white points chosen in both the reference sensor color space
and the CIE XYZ colorspace when calculating this transform will
match the standard white point for the second reference illuminant
(i.e. no chromatic adaptation will be applied by this transform).
This matrix will only be present if the second reference
illuminant is present.
</details>
<tag id="RAW" />
</entry>
<entry name="forwardMatrix1" type="rational"
visibility="public" optional="true"
type_notes="3x3 matrix in row-major-order" container="array"
typedef="colorSpaceTransform">
<array>
<size>3</size>
<size>3</size>
</array>
<description>
A matrix that transforms white balanced camera colors from the reference
sensor colorspace to the CIE XYZ colorspace with a D50 whitepoint.
</description>
<details>
This matrix is used to convert to the standard CIE XYZ colorspace, and
is used when processing raw buffer data.
This matrix is expressed as a 3x3 matrix in row-major-order, and contains
a color transform matrix that maps white balanced colors from the
reference sensor color space to the CIE XYZ color space with a D50 white
point.
Under the first reference illuminant (android.sensor.referenceIlluminant1)
this matrix is chosen so that the standard white point for this reference
illuminant in the reference sensor colorspace is mapped to D50 in the
CIE XYZ colorspace.
</details>
<tag id="RAW" />
</entry>
<entry name="forwardMatrix2" type="rational"
visibility="public" optional="true"
type_notes="3x3 matrix in row-major-order" container="array"
typedef="colorSpaceTransform">
<array>
<size>3</size>
<size>3</size>
</array>
<description>
A matrix that transforms white balanced camera colors from the reference
sensor colorspace to the CIE XYZ colorspace with a D50 whitepoint.
</description>
<details>
This matrix is used to convert to the standard CIE XYZ colorspace, and
is used when processing raw buffer data.
This matrix is expressed as a 3x3 matrix in row-major-order, and contains
a color transform matrix that maps white balanced colors from the
reference sensor color space to the CIE XYZ color space with a D50 white
point.
Under the second reference illuminant (android.sensor.referenceIlluminant2)
this matrix is chosen so that the standard white point for this reference
illuminant in the reference sensor colorspace is mapped to D50 in the
CIE XYZ colorspace.
This matrix will only be present if the second reference
illuminant is present.
</details>
<tag id="RAW" />
</entry>
<entry name="baseGainFactor" type="rational"
optional="true">
<description>Gain factor from electrons to raw units when
ISO=100</description>
<tag id="FUTURE" />
</entry>
<entry name="blackLevelPattern" type="int32" visibility="public"
optional="true" type_notes="2x2 raw count block" container="array"
typedef="blackLevelPattern">
<array>
<size>4</size>
</array>
<description>
A fixed black level offset for each of the color filter arrangement
(CFA) mosaic channels.
</description>
<range>&amp;gt;= 0 for each.</range>
<details>
This tag specifies the zero light value for each of the CFA mosaic
channels in the camera sensor. The maximal value output by the
sensor is represented by the value in android.sensor.info.whiteLevel.
The values are given in the same order as channels listed for the CFA
layout tag (see android.sensor.info.colorFilterArrangement), i.e. the
nth value given corresponds to the black level offset for the nth
color channel listed in the CFA.
</details>
<hal_details>
The values are given in row-column scan order, with the first value
corresponding to the element of the CFA in row=0, column=0.
</hal_details>
<tag id="RAW" />
</entry>
<entry name="maxAnalogSensitivity" type="int32" visibility="public"
optional="true" hwlevel="full">
<description>Maximum sensitivity that is implemented
purely through analog gain.</description>
<details>For android.sensor.sensitivity values less than or
equal to this, all applied gain must be analog. For
values above this, the gain applied can be a mix of analog and
digital.</details>
<tag id="V1" />
<tag id="FULL" />
</entry>
<entry name="orientation" type="int32" visibility="public"
hwlevel="legacy">
<description>Clockwise angle through which the output
image needs to be rotated to be upright on the device
screen in its native orientation. Also defines the
direction of rolling shutter readout, which is from top
to bottom in the sensor's coordinate system</description>
<units>degrees clockwise rotation, only multiples of
90</units>
<range>0,90,180,270</range>
<tag id="BC" />
</entry>
<entry name="profileHueSatMapDimensions" type="int32"
visibility="system" optional="true"
type_notes="Number of samples for hue, saturation, and value"
container="array">
<array>
<size>3</size>
</array>
<description>
The number of input samples for each dimension of
android.sensor.profileHueSatMap.
</description>
<range>
Hue &amp;gt;= 1,
Saturation &amp;gt;= 2,
Value &amp;gt;= 1
</range>
<details>
The number of input samples for the hue, saturation, and value
dimension of android.sensor.profileHueSatMap. The order of the
dimensions given is hue, saturation, value; where hue is the 0th
element.
</details>
<tag id="RAW" />
</entry>
</static>
<dynamic>
<clone entry="android.sensor.exposureTime" kind="controls">
</clone>
<clone entry="android.sensor.frameDuration"
kind="controls"></clone>
<clone entry="android.sensor.sensitivity" kind="controls">
</clone>
<entry name="timestamp" type="int64" visibility="public"
hwlevel="legacy">
<description>Time at start of exposure of first
row of the image sensor active array, in nanoseconds.</description>
<units>nanoseconds</units>
<range>&amp;gt; 0</range>
<details>The timestamps are also included in all image
buffers produced for the same capture, and will be identical
on all the outputs.
When android.sensor.info.timestampSource `==` UNKNOWN,
the timestamps measure time since an unspecified starting point,
and are monotonically increasing. They can be compared with the
timestamps for other captures from the same camera device, but are
not guaranteed to be comparable to any other time source.
When android.sensor.info.timestampSource `==` REALTIME,
the timestamps measure time in the same timebase as
android.os.SystemClock#elapsedRealtimeNanos(), and they can be
compared to other timestamps from other subsystems that are using
that base.
</details>
<hal_details>
All timestamps must be in reference to the kernel's
CLOCK_BOOTTIME monotonic clock, which properly accounts for
time spent asleep. This allows for synchronization with
sensors that continue to operate while the system is
otherwise asleep.
If android.sensor.info.timestampSource `==` REALTIME,
The timestamp must be synchronized with the timestamps from other
sensor subsystems that are using the same timebase.
</hal_details>
<tag id="BC" />
</entry>
<entry name="temperature" type="float"
optional="true">
<description>The temperature of the sensor, sampled at the time
exposure began for this frame.
The thermal diode being queried should be inside the sensor PCB, or
somewhere close to it.
</description>
<units>celsius</units>
<range>Optional. This value is missing if no temperature is available.</range>
<tag id="FUTURE" />
</entry>
<entry name="neutralColorPoint" type="rational" visibility="public"
optional="true" container="array">
<array>
<size>3</size>
</array>
<description>
The estimated camera neutral color in the native sensor colorspace at
the time of capture.
</description>
<details>
This value gives the neutral color point encoded as an RGB value in the
native sensor color space. The neutral color point indicates the
currently estimated white point of the scene illumination. It can be
used to interpolate between the provided color transforms when
processing raw sensor data.
The order of the values is R, G, B; where R is in the lowest index.
</details>
<tag id="RAW" />
</entry>
<entry name="noiseProfile" type="double" visibility="public"
optional="true" type_notes="Pairs of noise model coefficients"
container="array" typedef="pairDoubleDouble">
<array>
<size>2</size>
<size>CFA Channels</size>
</array>
<description>
Noise model coefficients for each CFA mosaic channel.
</description>
<details>
This tag contains two noise model coefficients for each CFA channel
corresponding to the sensor amplification (S) and sensor readout
noise (O). These are given as pairs of coefficients for each channel
in the same order as channels listed for the CFA layout tag
(see android.sensor.info.colorFilterArrangement). This is
represented as an array of Pair&amp;lt;Double, Double&amp;gt;, where
the first member of the Pair at index n is the S coefficient and the
second member is the O coefficient for the nth color channel in the CFA.
These coefficients are used in a two parameter noise model to describe
the amount of noise present in the image for each CFA channel. The
noise model used here is:
N(x) = sqrt(Sx + O)
Where x represents the recorded signal of a CFA channel normalized to
the range [0, 1], and S and O are the noise model coeffiecients for
that channel.
A more detailed description of the noise model can be found in the
Adobe DNG specification for the NoiseProfile tag.
</details>
<hal_details>
For a CFA layout of RGGB, the list of coefficients would be given as
an array of doubles S0,O0,S1,O1,..., where S0 and O0 are the coefficients
for the red channel, S1 and O1 are the coefficients for the first green
channel, etc.
</hal_details>
<tag id="RAW" />
</entry>
<entry name="profileHueSatMap" type="float"
visibility="system" optional="true"
type_notes="Mapping for hue, saturation, and value"
container="array">
<array>
<size>hue_samples</size>
<size>saturation_samples</size>
<size>value_samples</size>
<size>3</size>
</array>
<description>
A mapping containing a hue shift, saturation scale, and value scale
for each pixel.
</description>
<units>
Hue shift is given in degrees; saturation and value scale factors are
unitless.
</units>
<details>
hue_samples, saturation_samples, and value_samples are given in
android.sensor.profileHueSatMapDimensions.
Each entry of this map contains three floats corresponding to the
hue shift, saturation scale, and value scale, respectively; where the
hue shift has the lowest index. The map entries are stored in the tag
in nested loop order, with the value divisions in the outer loop, the
hue divisions in the middle loop, and the saturation divisions in the
inner loop. All zero input saturation entries are required to have a
value scale factor of 1.0.
</details>
<tag id="RAW" />
</entry>
<entry name="profileToneCurve" type="float"
visibility="system" optional="true"
type_notes="Samples defining a spline for a tone-mapping curve"
container="array">
<array>
<size>samples</size>
<size>2</size>
</array>
<description>
A list of x,y samples defining a tone-mapping curve for gamma adjustment.
</description>
<range>
Each sample has an input range of `[0, 1]` and an output range of
`[0, 1]`. The first sample is required to be `(0, 0)`, and the last
sample is required to be `(1, 1)`.
</range>
<details>
This tag contains a default tone curve that can be applied while
processing the image as a starting point for user adjustments.
The curve is specified as a list of value pairs in linear gamma.
The curve is interpolated using a cubic spline.
</details>
<tag id="RAW" />
</entry>
<entry name="greenSplit" type="float" visibility="public" optional="true">
<description>
The worst-case divergence between Bayer green channels.
</description>
<range>
&amp;gt;= 0
</range>
<details>
This value is an estimate of the worst case split between the
Bayer green channels in the red and blue rows in the sensor color
filter array.
The green split is calculated as follows:
1. A 5x5 pixel (or larger) window W within the active sensor array is
chosen. The term 'pixel' here is taken to mean a group of 4 Bayer
mosaic channels (R, Gr, Gb, B). The location and size of the window
chosen is implementation defined, and should be chosen to provide a
green split estimate that is both representative of the entire image
for this camera sensor, and can be calculated quickly.
1. The arithmetic mean of the green channels from the red
rows (mean_Gr) within W is computed.
1. The arithmetic mean of the green channels from the blue
rows (mean_Gb) within W is computed.
1. The maximum ratio R of the two means is computed as follows:
`R = max((mean_Gr + 1)/(mean_Gb + 1), (mean_Gb + 1)/(mean_Gr + 1))`
The ratio R is the green split divergence reported for this property,
which represents how much the green channels differ in the mosaic
pattern. This value is typically used to determine the treatment of
the green mosaic channels when demosaicing.
The green split value can be roughly interpreted as follows:
* R &amp;lt; 1.03 is a negligible split (&amp;lt;3% divergence).
* 1.20 &amp;lt;= R &amp;gt;= 1.03 will require some software
correction to avoid demosaic errors (3-20% divergence).
* R &amp;gt; 1.20 will require strong software correction to produce
a usuable image (&amp;gt;20% divergence).
</details>
<hal_details>
The green split given may be a static value based on prior
characterization of the camera sensor using the green split
calculation method given here over a large, representative, sample
set of images. Other methods of calculation that produce equivalent
results, and can be interpreted in the same manner, may be used.
</hal_details>
<tag id="RAW" />
</entry>
</dynamic>
<controls>
<entry name="testPatternData" type="int32" visibility="public" optional="true" container="array">
<array>
<size>4</size>
</array>
<description>
A pixel `[R, G_even, G_odd, B]` that supplies the test pattern
when android.sensor.testPatternMode is SOLID_COLOR.
</description>
<range>Optional.
Must be supported if android.sensor.availableTestPatternModes contains
SOLID_COLOR.</range>
<details>
Each color channel is treated as an unsigned 32-bit integer.
The camera device then uses the most significant X bits
that correspond to how many bits are in its Bayer raw sensor
output.
For example, a sensor with RAW10 Bayer output would use the
10 most significant bits from each color channel.
</details>
<hal_details>
</hal_details>
</entry>
<entry name="testPatternMode" type="int32" visibility="public" optional="true"
enum="true">
<enum>
<value>OFF
<notes>No test pattern mode is used, and the camera
device returns captures from the image sensor.
This is the default if the key is not set.</notes>
</value>
<value>SOLID_COLOR
<notes>
Each pixel in `[R, G_even, G_odd, B]` is replaced by its
respective color channel provided in
android.sensor.testPatternData.
For example:
android.testPatternData = [0, 0xFFFFFFFF, 0xFFFFFFFF, 0]
All green pixels are 100% green. All red/blue pixels are black.
android.testPatternData = [0xFFFFFFFF, 0, 0xFFFFFFFF, 0]
All red pixels are 100% red. Only the odd green pixels
are 100% green. All blue pixels are 100% black.
</notes>
</value>
<value>COLOR_BARS
<notes>
All pixel data is replaced with an 8-bar color pattern.
The vertical bars (left-to-right) are as follows:
* 100% white
* yellow
* cyan
* green
* magenta
* red
* blue
* black
In general the image would look like the following:
W Y C G M R B K
W Y C G M R B K
W Y C G M R B K
W Y C G M R B K
W Y C G M R B K
. . . . . . . .
. . . . . . . .
. . . . . . . .
(B = Blue, K = Black)
Each bar should take up 1/8 of the sensor pixel array width.
When this is not possible, the bar size should be rounded
down to the nearest integer and the pattern can repeat
on the right side.
Each bar's height must always take up the full sensor
pixel array height.
Each pixel in this test pattern must be set to either
0% intensity or 100% intensity.
</notes>
</value>
<value>COLOR_BARS_FADE_TO_GRAY
<notes>
The test pattern is similar to COLOR_BARS, except that
each bar should start at its specified color at the top,
and fade to gray at the bottom.
Furthermore each bar is further subdivided into a left and
right half. The left half should have a smooth gradient,
and the right half should have a quantized gradient.
In particular, the right half's should consist of blocks of the
same color for 1/16th active sensor pixel array width.
The least significant bits in the quantized gradient should
be copied from the most significant bits of the smooth gradient.
The height of each bar should always be a multiple of 128.
When this is not the case, the pattern should repeat at the bottom
of the image.
</notes>
</value>
<value>PN9
<notes>
All pixel data is replaced by a pseudo-random sequence
generated from a PN9 512-bit sequence (typically implemented
in hardware with a linear feedback shift register).
The generator should be reset at the beginning of each frame,
and thus each subsequent raw frame with this test pattern should
be exactly the same as the last.
</notes>
</value>
<value id="256">CUSTOM1
<notes>The first custom test pattern. All custom patterns that are
available only on this camera device are at least this numeric
value.
All of the custom test patterns will be static
(that is the raw image must not vary from frame to frame).
</notes>
</value>
</enum>
<description>When enabled, the sensor sends a test pattern instead of
doing a real exposure from the camera.
</description>
<range>Optional. Defaults to OFF. Value must be one of
android.sensor.availableTestPatternModes</range>
<details>
When a test pattern is enabled, all manual sensor controls specified
by android.sensor.* will be ignored. All other controls should
work as normal.
For example, if manual flash is enabled, flash firing should still
occur (and that the test pattern remain unmodified, since the flash
would not actually affect it).
</details>
<hal_details>
All test patterns are specified in the Bayer domain.
The HAL may choose to substitute test patterns from the sensor
with test patterns from on-device memory. In that case, it should be
indistinguishable to the ISP whether the data came from the
sensor interconnect bus (such as CSI2) or memory.
</hal_details>
</entry>
</controls>
<dynamic>
<clone entry="android.sensor.testPatternData" kind="controls">
</clone>
<clone entry="android.sensor.testPatternMode" kind="controls">
</clone>
</dynamic>
<static>
<entry name="availableTestPatternModes" type="int32" visibility="public" optional="true"
type_notes="list of enums" container="array">
<array>
<size>n</size>
</array>
<description>Lists the supported sensor test pattern modes for android.sensor.testPatternMode.
</description>
<range>Always includes OFF if defined. All custom modes must be >= CUSTOM1</range>
<details>
Optional. Defaults to [OFF].
</details>
</entry>
</static>
<dynamic>
<entry name="rollingShutterSkew" type="int64" visibility="public" hwlevel="limited">
<description>Duration between the start of first row exposure
and the start of last row exposure.</description>
<units>nanoseconds</units>
<range> &amp;gt;= 0 and &amp;lt;
StreamConfigurationMap#getOutputMinFrameDuration(int, Size).</range>
<details>
This is the exposure time skew (in the unit of nanosecond) between the first and
last row exposure start times. The first row and the last row are the first
and last rows inside of the android.sensor.info.activeArraySize.
For typical camera sensors that use rolling shutters, this is also equivalent
to the frame readout time.
</details>
<hal_details>
The HAL must report `0` if the sensor is using global shutter, where all pixels begin
exposure at the same time.
</hal_details>
<tag id="V1" />
</entry>
</dynamic>
</section>
<section name="shading">
<controls>
<entry name="mode" type="byte" visibility="public" enum="true" hwlevel="full">
<enum>
<value>OFF
<notes>No lens shading correction is applied.</notes></value>
<value>FAST
<notes>Apply lens shading corrections, without slowing
frame rate relative to sensor raw output</notes></value>
<value>HIGH_QUALITY
<notes>Apply high-quality lens shading correction, at the
cost of reduced frame rate.</notes></value>
</enum>
<description>Quality of lens shading correction applied
to the image data.</description>
<details>
When set to OFF mode, no lens shading correction will be applied by the
camera device, and an identity lens shading map data will be provided
if `android.statistics.lensShadingMapMode == ON`. For example, for lens
shading map with size specified as `android.lens.info.shadingMapSize = [ 4, 3 ]`,
the output android.statistics.lensShadingMap for this case will be an identity map
shown below:
[ 1.0, 1.0, 1.0, 1.0, 1.0, 1.0, 1.0, 1.0,
1.0, 1.0, 1.0, 1.0, 1.0, 1.0, 1.0, 1.0,
1.0, 1.0, 1.0, 1.0, 1.0, 1.0, 1.0, 1.0,
1.0, 1.0, 1.0, 1.0, 1.0, 1.0, 1.0, 1.0,
1.0, 1.0, 1.0, 1.0, 1.0, 1.0, 1.0, 1.0,
1.0, 1.0, 1.0, 1.0, 1.0, 1.0, 1.0, 1.0 ]
When set to other modes, lens shading correction will be applied by the
camera device. Applications can request lens shading map data by setting
android.statistics.lensShadingMapMode to ON, and then the camera device will provide
lens shading map data in android.statistics.lensShadingMap, with size specified
by android.lens.info.shadingMapSize; the returned shading map data will be the one
applied by the camera device for this capture request.
The shading map data may depend on the auto-exposure (AE) and AWB statistics, therefore the reliability
of the map data may be affected by the AE and AWB algorithms. When AE and AWB are in
AUTO modes(android.control.aeMode `!=` OFF and android.control.awbMode `!=` OFF),
to get best results, it is recommended that the applications wait for the AE and AWB to
be converged before using the returned shading map data.
</details>
</entry>
<entry name="strength" type="byte">
<description>Control the amount of shading correction
applied to the images</description>
<units>unitless: 1-10; 10 is full shading
compensation</units>
<tag id="FUTURE" />
</entry>
</controls>
<dynamic>
<clone entry="android.shading.mode" kind="controls">
</clone>
</dynamic>
</section>
<section name="statistics">
<controls>
<entry name="faceDetectMode" type="byte" visibility="public" enum="true"
hwlevel="legacy">
<enum>
<value>OFF
<notes>Do not include face detection statistics in capture
results.</notes></value>
<value optional="true">SIMPLE
<notes>Return face rectangle and confidence values only.
In this mode, only android.statistics.faceRectangles and
android.statistics.faceScores outputs are valid.
</notes></value>
<value optional="true">FULL
<notes>Return all face
metadata.
In this mode,
android.statistics.faceRectangles,
android.statistics.faceScores,
android.statistics.faceIds, and
android.statistics.faceLandmarks outputs are valid.
</notes></value>
</enum>
<description>Control for the face detector
unit.</description>
<range>
android.statistics.info.availableFaceDetectModes</range>
<details>Whether face detection is enabled, and whether it
should output just the basic fields or the full set of
fields. Value must be one of the
android.statistics.info.availableFaceDetectModes.</details>
<tag id="BC" />
</entry>
<entry name="histogramMode" type="byte" enum="true" typedef="boolean">
<enum>
<value>OFF</value>
<value>ON</value>
</enum>
<description>Operating mode for histogram
generation</description>
<tag id="FUTURE" />
</entry>
<entry name="sharpnessMapMode" type="byte" enum="true" typedef="boolean">
<enum>
<value>OFF</value>
<value>ON</value>
</enum>
<description>Operating mode for sharpness map
generation</description>
<tag id="FUTURE" />
</entry>
<entry name="hotPixelMapMode" type="byte" visibility="public" enum="true"
typedef="boolean">
<enum>
<value>OFF
<notes>Hot pixel map production is disabled.
</notes></value>
<value>ON
<notes>Hot pixel map production is enabled.
</notes></value>
</enum>
<description>
Operating mode for hotpixel map generation.
</description>
<details>
If set to ON, a hotpixel map is returned in android.statistics.hotPixelMap.
If set to OFF, no hotpixel map will be returned.
This must be set to a valid mode from android.statistics.info.availableHotPixelMapModes.
</details>
<tag id="V1" />
<tag id="RAW" />
</entry>
</controls>
<static>
<namespace name="info">
<entry name="availableFaceDetectModes" type="byte"
visibility="public"
type_notes="List of enums from android.statistics.faceDetectMode"
container="array"
typedef="enumList"
hwlevel="legacy">
<array>
<size>n</size>
</array>
<description>The face detection modes that are available
for this camera device.
</description>
<units>List of enum:
OFF
SIMPLE
FULL</units>
<details>OFF is always supported.
SIMPLE means the device supports the
android.statistics.faceRectangles and
android.statistics.faceScores outputs.
FULL means the device additionally supports the
android.statistics.faceIds and
android.statistics.faceLandmarks outputs.
</details>
</entry>
<entry name="histogramBucketCount" type="int32">
<description>Number of histogram buckets
supported</description>
<range>&amp;gt;= 64</range>
<tag id="FUTURE" />
</entry>
<entry name="maxFaceCount" type="int32" visibility="public" hwlevel="legacy">
<description>The maximum number of simultaneously detectable
faces.</description>
<range>0 for cameras without available face detection; otherwise:
`&gt;=4` for LIMITED or FULL hwlevel devices or
`&gt;0` for LEGACY devices.</range>
<tag id="BC" />
</entry>
<entry name="maxHistogramCount" type="int32">
<description>Maximum value possible for a histogram
bucket</description>
<tag id="FUTURE" />
</entry>
<entry name="maxSharpnessMapValue" type="int32">
<description>Maximum value possible for a sharpness map
region.</description>
<tag id="FUTURE" />
</entry>
<entry name="sharpnessMapSize" type="int32"
type_notes="width x height" container="array" typedef="size">
<array>
<size>2</size>
</array>
<description>Dimensions of the sharpness
map</description>
<range>Must be at least 32 x 32</range>
<tag id="FUTURE" />
</entry>
<entry name="availableHotPixelMapModes" type="byte" visibility="public"
type_notes="list of enums" container="array" typedef="boolean">
<array>
<size>n</size>
</array>
<description>
The set of hot pixel map output modes supported by this camera device.
</description>
<details>
This tag lists valid output modes for android.statistics.hotPixelMapMode.
If no hotpixel map is available for this camera device, this will contain
only OFF. If the hotpixel map is available, this will include both
the ON and OFF options.
Required on devices with the RAW capability.
</details>
<tag id="V1" />
<tag id="RAW" />
</entry>
</namespace>
</static>
<dynamic>
<clone entry="android.statistics.faceDetectMode"
kind="controls"></clone>
<entry name="faceIds" type="int32" visibility="hidden" container="array"
hwlevel="legacy">
<array>
<size>n</size>
</array>
<description>List of unique IDs for detected faces.</description>
<details>
Each detected face is given a unique ID that is valid for as long as the face is visible
to the camera device. A face that leaves the field of view and later returns may be
assigned a new ID.
Only available if android.statistics.faceDetectMode == FULL</details>
<tag id="BC" />
</entry>
<entry name="faceLandmarks" type="int32" visibility="hidden"
type_notes="(leftEyeX, leftEyeY, rightEyeX, rightEyeY, mouthX, mouthY)"
container="array" hwlevel="legacy">
<array>
<size>n</size>
<size>6</size>
</array>
<description>List of landmarks for detected
faces.</description>
<details>
The coordinate system is that of android.sensor.info.activeArraySize, with
`(0, 0)` being the top-left pixel of the active array.
Only available if android.statistics.faceDetectMode == FULL</details>
<tag id="BC" />
</entry>
<entry name="faceRectangles" type="int32" visibility="hidden"
type_notes="(xmin, ymin, xmax, ymax). (0,0) is top-left of active pixel area"
container="array" typedef="rectangle" hwlevel="legacy">
<array>
<size>n</size>
<size>4</size>
</array>
<description>List of the bounding rectangles for detected
faces.</description>
<details>
The coordinate system is that of android.sensor.info.activeArraySize, with
`(0, 0)` being the top-left pixel of the active array.
Only available if android.statistics.faceDetectMode != OFF</details>
<tag id="BC" />
</entry>
<entry name="faceScores" type="byte" visibility="hidden" container="array"
hwlevel="legacy">
<array>
<size>n</size>
</array>
<description>List of the face confidence scores for
detected faces</description>
<range>1-100</range>
<details>Only available if android.statistics.faceDetectMode != OFF.
</details>
<hal_details>
The value should be meaningful (for example, setting 100 at
all times is illegal).</hal_details>
<tag id="BC" />
</entry>
<entry name="faces" type="int32" visibility="public" synthetic="true"
container="array" typedef="face" hwlevel="legacy">
<array>
<size>n</size>
</array>
<description>List of the faces detected through camera face detection
in this result.</description>
<details>
Only available if android.statistics.faceDetectMode `!=` OFF.
</details>
</entry>
<entry name="histogram" type="int32"
type_notes="count of pixels for each color channel that fall into each histogram bucket, scaled to be between 0 and maxHistogramCount"
container="array">
<array>
<size>n</size>
<size>3</size>
</array>
<description>A 3-channel histogram based on the raw
sensor data</description>
<details>The k'th bucket (0-based) covers the input range
(with w = android.sensor.info.whiteLevel) of [ k * w/N,
(k + 1) * w / N ). If only a monochrome sharpness map is
supported, all channels should have the same data</details>
<tag id="FUTURE" />
</entry>
<clone entry="android.statistics.histogramMode"
kind="controls"></clone>
<entry name="sharpnessMap" type="int32"
type_notes="estimated sharpness for each region of the input image. Normalized to be between 0 and maxSharpnessMapValue. Higher values mean sharper (better focused)"
container="array">
<array>
<size>n</size>
<size>m</size>
<size>3</size>
</array>
<description>A 3-channel sharpness map, based on the raw
sensor data</description>
<details>If only a monochrome sharpness map is supported,
all channels should have the same data</details>
<tag id="FUTURE" />
</entry>
<clone entry="android.statistics.sharpnessMapMode"
kind="controls"></clone>
<entry name="lensShadingCorrectionMap" type="byte" visibility="public"
typedef="lensShadingMap" hwlevel="full">
<description>The shading map is a low-resolution floating-point map
that lists the coefficients used to correct for vignetting, for each
Bayer color channel.</description>
<range>Each gain factor is &amp;gt;= 1</range>
<details>The least shaded section of the image should have a gain factor
of 1; all other sections should have gains above 1.
When android.colorCorrection.mode = TRANSFORM_MATRIX, the map
must take into account the colorCorrection settings.
The shading map is for the entire active pixel array, and is not
affected by the crop region specified in the request. Each shading map
entry is the value of the shading compensation map over a specific
pixel on the sensor. Specifically, with a (N x M) resolution shading
map, and an active pixel array size (W x H), shading map entry
(x,y) ϵ (0 ... N-1, 0 ... M-1) is the value of the shading map at
pixel ( ((W-1)/(N-1)) * x, ((H-1)/(M-1)) * y) for the four color channels.
The map is assumed to be bilinearly interpolated between the sample points.
The channel order is [R, Geven, Godd, B], where Geven is the green
channel for the even rows of a Bayer pattern, and Godd is the odd rows.
The shading map is stored in a fully interleaved format.
The shading map should have on the order of 30-40 rows and columns,
and must be smaller than 64x64.
As an example, given a very small map defined as:
width,height = [ 4, 3 ]
values =
[ 1.3, 1.2, 1.15, 1.2, 1.2, 1.2, 1.15, 1.2,
1.1, 1.2, 1.2, 1.2, 1.3, 1.2, 1.3, 1.3,
1.2, 1.2, 1.25, 1.1, 1.1, 1.1, 1.1, 1.0,
1.0, 1.0, 1.0, 1.0, 1.2, 1.3, 1.25, 1.2,
1.3, 1.2, 1.2, 1.3, 1.2, 1.15, 1.1, 1.2,
1.2, 1.1, 1.0, 1.2, 1.3, 1.15, 1.2, 1.3 ]
The low-resolution scaling map images for each channel are
(displayed using nearest-neighbor interpolation):
![Red lens shading map](android.statistics.lensShadingMap/red_shading.png)
![Green (even rows) lens shading map](android.statistics.lensShadingMap/green_e_shading.png)
![Green (odd rows) lens shading map](android.statistics.lensShadingMap/green_o_shading.png)
![Blue lens shading map](android.statistics.lensShadingMap/blue_shading.png)
As a visualization only, inverting the full-color map to recover an
image of a gray wall (using bicubic interpolation for visual quality) as captured by the sensor gives:
![Image of a uniform white wall (inverse shading map)](android.statistics.lensShadingMap/inv_shading.png)
</details>
</entry>
<entry name="lensShadingMap" type="float" visibility="hidden"
type_notes="2D array of float gain factors per channel to correct lens shading"
container="array" hwlevel="full">
<array>
<size>4</size>
<size>n</size>
<size>m</size>
</array>
<description>The shading map is a low-resolution floating-point map
that lists the coefficients used to correct for vignetting, for each
Bayer color channel.</description>
<range>Each gain factor is &amp;gt;= 1</range>
<details>The least shaded section of the image should have a gain factor
of 1; all other sections should have gains above 1.
When android.colorCorrection.mode = TRANSFORM_MATRIX, the map
must take into account the colorCorrection settings.
The shading map is for the entire active pixel array, and is not
affected by the crop region specified in the request. Each shading map
entry is the value of the shading compensation map over a specific
pixel on the sensor. Specifically, with a (N x M) resolution shading
map, and an active pixel array size (W x H), shading map entry
(x,y) ϵ (0 ... N-1, 0 ... M-1) is the value of the shading map at
pixel ( ((W-1)/(N-1)) * x, ((H-1)/(M-1)) * y) for the four color channels.
The map is assumed to be bilinearly interpolated between the sample points.
The channel order is [R, Geven, Godd, B], where Geven is the green
channel for the even rows of a Bayer pattern, and Godd is the odd rows.
The shading map is stored in a fully interleaved format, and its size
is provided in the camera static metadata by android.lens.info.shadingMapSize.
The shading map should have on the order of 30-40 rows and columns,
and must be smaller than 64x64.
As an example, given a very small map defined as:
android.lens.info.shadingMapSize = [ 4, 3 ]
android.statistics.lensShadingMap =
[ 1.3, 1.2, 1.15, 1.2, 1.2, 1.2, 1.15, 1.2,
1.1, 1.2, 1.2, 1.2, 1.3, 1.2, 1.3, 1.3,
1.2, 1.2, 1.25, 1.1, 1.1, 1.1, 1.1, 1.0,
1.0, 1.0, 1.0, 1.0, 1.2, 1.3, 1.25, 1.2,
1.3, 1.2, 1.2, 1.3, 1.2, 1.15, 1.1, 1.2,
1.2, 1.1, 1.0, 1.2, 1.3, 1.15, 1.2, 1.3 ]
The low-resolution scaling map images for each channel are
(displayed using nearest-neighbor interpolation):
![Red lens shading map](android.statistics.lensShadingMap/red_shading.png)
![Green (even rows) lens shading map](android.statistics.lensShadingMap/green_e_shading.png)
![Green (odd rows) lens shading map](android.statistics.lensShadingMap/green_o_shading.png)
![Blue lens shading map](android.statistics.lensShadingMap/blue_shading.png)
As a visualization only, inverting the full-color map to recover an
image of a gray wall (using bicubic interpolation for visual quality) as captured by the sensor gives:
![Image of a uniform white wall (inverse shading map)](android.statistics.lensShadingMap/inv_shading.png)
</details>
<hal_details>
The lens shading map calculation may depend on exposure and white balance statistics.
When AE and AWB are in AUTO modes
(android.control.aeMode `!=` OFF and android.control.awbMode `!=` OFF), the HAL
may have all the information it need to generate most accurate lens shading map. When
AE or AWB are in manual mode
(android.control.aeMode `==` OFF or android.control.awbMode `==` OFF), the shading map
may be adversely impacted by manual exposure or white balance parameters. To avoid
generating unreliable shading map data, the HAL may choose to lock the shading map with
the latest known good map generated when the AE and AWB are in AUTO modes.
</hal_details>
</entry>
<entry name="predictedColorGains" type="float"
visibility="hidden"
deprecated="true"
optional="true"
type_notes="A 1D array of floats for 4 color channel gains"
container="array">
<array>
<size>4</size>
</array>
<description>The best-fit color channel gains calculated
by the camera device's statistics units for the current output frame.
</description>
<details>
This may be different than the gains used for this frame,
since statistics processing on data from a new frame
typically completes after the transform has already been
applied to that frame.
The 4 channel gains are defined in Bayer domain,
see android.colorCorrection.gains for details.
This value should always be calculated by the auto-white balance (AWB) block,
regardless of the android.control.* current values.
</details>
</entry>
<entry name="predictedColorTransform" type="rational"
visibility="hidden"
deprecated="true"
optional="true"
type_notes="3x3 rational matrix in row-major order"
container="array">
<array>
<size>3</size>
<size>3</size>
</array>
<description>The best-fit color transform matrix estimate
calculated by the camera device's statistics units for the current
output frame.</description>
<details>The camera device will provide the estimate from its
statistics unit on the white balance transforms to use
for the next frame. These are the values the camera device believes
are the best fit for the current output frame. This may
be different than the transform used for this frame, since
statistics processing on data from a new frame typically
completes after the transform has already been applied to
that frame.
These estimates must be provided for all frames, even if
capture settings and color transforms are set by the application.
This value should always be calculated by the auto-white balance (AWB) block,
regardless of the android.control.* current values.
</details>
</entry>
<entry name="sceneFlicker" type="byte" visibility="public" enum="true"
hwlevel="full">
<enum>
<value>NONE
<notes>The camera device does not detect any flickering illumination
in the current scene.</notes></value>
<value>50HZ
<notes>The camera device detects illumination flickering at 50Hz
in the current scene.</notes></value>
<value>60HZ
<notes>The camera device detects illumination flickering at 60Hz
in the current scene.</notes></value>
</enum>
<description>The camera device estimated scene illumination lighting
frequency.</description>
<details>
Many light sources, such as most fluorescent lights, flicker at a rate
that depends on the local utility power standards. This flicker must be
accounted for by auto-exposure routines to avoid artifacts in captured images.
The camera device uses this entry to tell the application what the scene
illuminant frequency is.
When manual exposure control is enabled
(`android.control.aeMode == OFF` or `android.control.mode ==
OFF`), the android.control.aeAntibandingMode doesn't perform
antibanding, and the application can ensure it selects
exposure times that do not cause banding issues by looking
into this metadata field. See
android.control.aeAntibandingMode for more details.
Reports NONE if there doesn't appear to be flickering illumination.
</details>
</entry>
<clone entry="android.statistics.hotPixelMapMode" kind="controls">
</clone>
<entry name="hotPixelMap" type="int32" visibility="public"
type_notes="list of coordinates based on android.sensor.pixelArraySize"
container="array" typedef="point">
<array>
<size>2</size>
<size>n</size>
</array>
<description>
List of `(x, y)` coordinates of hot/defective pixels on the sensor.
</description>
<range>
n &lt;= number of pixels on the sensor.
The `(x, y)` coordinates must be bounded by
android.sensor.info.pixelArraySize.
</range>
<details>
A coordinate `(x, y)` must lie between `(0, 0)`, and
`(width - 1, height - 1)` (inclusive), which are the top-left and
bottom-right of the pixel array, respectively. The width and
height dimensions are given in android.sensor.info.pixelArraySize.
This may include hot pixels that lie outside of the active array
bounds given by android.sensor.info.activeArraySize.
</details>
<hal_details>
A hotpixel map contains the coordinates of pixels on the camera
sensor that do report valid values (usually due to defects in
the camera sensor). This includes pixels that are stuck at certain
values, or have a response that does not accuractly encode the
incoming light from the scene.
To avoid performance issues, there should be significantly fewer hot
pixels than actual pixels on the camera sensor.
</hal_details>
<tag id="V1" />
<tag id="RAW" />
</entry>
</dynamic>
<controls>
<entry name="lensShadingMapMode" type="byte" visibility="public" enum="true" hwlevel="full">
<enum>
<value>OFF
<notes>Do not include a lens shading map in the capture result.</notes></value>
<value>ON
<notes>Include a lens shading map in the capture result.</notes></value>
</enum>
<description>Whether the camera device will output the lens
shading map in output result metadata.</description>
<details>When set to ON,
android.statistics.lensShadingMap will be provided in
the output result metadata.
Required for devices with the RAW capability.
</details>
<tag id="RAW" />
</entry>
</controls>
<dynamic>
<clone entry="android.statistics.lensShadingMapMode" kind="controls">
</clone>
</dynamic>
</section>
<section name="tonemap">
<controls>
<entry name="curveBlue" type="float" visibility="hidden"
type_notes="1D array of float pairs (P_IN, P_OUT). The maximum number of pairs is specified by android.tonemap.maxCurvePoints."
container="array" hwlevel="full">
<array>
<size>n</size>
<size>2</size>
</array>
<description>Tonemapping / contrast / gamma curve for the blue
channel, to use when android.tonemap.mode is
CONTRAST_CURVE.</description>
<units>same as android.tonemap.curveRed</units>
<range>same as android.tonemap.curveRed</range>
<details>See android.tonemap.curveRed for more details.</details>
</entry>
<entry name="curveGreen" type="float" visibility="hidden"
type_notes="1D array of float pairs (P_IN, P_OUT). The maximum number of pairs is specified by android.tonemap.maxCurvePoints."
container="array" hwlevel="full">
<array>
<size>n</size>
<size>2</size>
</array>
<description>Tonemapping / contrast / gamma curve for the green
channel, to use when android.tonemap.mode is
CONTRAST_CURVE.</description>
<units>same as android.tonemap.curveRed</units>
<range>same as android.tonemap.curveRed</range>
<details>See android.tonemap.curveRed for more details.</details>
</entry>
<entry name="curveRed" type="float" visibility="hidden"
type_notes="1D array of float pairs (P_IN, P_OUT). The maximum number of pairs is specified by android.tonemap.maxCurvePoints."
container="array" hwlevel="full">
<array>
<size>n</size>
<size>2</size>
</array>
<description>Tonemapping / contrast / gamma curve for the red
channel, to use when android.tonemap.mode is
CONTRAST_CURVE.</description>
<range>0-1 on both input and output coordinates, normalized
as a floating-point value such that 0 == black and 1 == white.
</range>
<details>
Each channel's curve is defined by an array of control points:
android.tonemap.curveRed =
[ P0in, P0out, P1in, P1out, P2in, P2out, P3in, P3out, ..., PNin, PNout ]
2 &lt;= N &lt;= android.tonemap.maxCurvePoints
These are sorted in order of increasing `Pin`; it is always
guaranteed that input values 0.0 and 1.0 are included in the list to
define a complete mapping. For input values between control points,
the camera device must linearly interpolate between the control
points.
Each curve can have an independent number of points, and the number
of points can be less than max (that is, the request doesn't have to
always provide a curve with number of points equivalent to
android.tonemap.maxCurvePoints).
A few examples, and their corresponding graphical mappings; these
only specify the red channel and the precision is limited to 4
digits, for conciseness.
Linear mapping:
android.tonemap.curveRed = [ 0, 0, 1.0, 1.0 ]
![Linear mapping curve](android.tonemap.curveRed/linear_tonemap.png)
Invert mapping:
android.tonemap.curveRed = [ 0, 1.0, 1.0, 0 ]
![Inverting mapping curve](android.tonemap.curveRed/inverse_tonemap.png)
Gamma 1/2.2 mapping, with 16 control points:
android.tonemap.curveRed = [
0.0000, 0.0000, 0.0667, 0.2920, 0.1333, 0.4002, 0.2000, 0.4812,
0.2667, 0.5484, 0.3333, 0.6069, 0.4000, 0.6594, 0.4667, 0.7072,
0.5333, 0.7515, 0.6000, 0.7928, 0.6667, 0.8317, 0.7333, 0.8685,
0.8000, 0.9035, 0.8667, 0.9370, 0.9333, 0.9691, 1.0000, 1.0000 ]
![Gamma = 1/2.2 tonemapping curve](android.tonemap.curveRed/gamma_tonemap.png)
Standard sRGB gamma mapping, per IEC 61966-2-1:1999, with 16 control points:
android.tonemap.curveRed = [
0.0000, 0.0000, 0.0667, 0.2864, 0.1333, 0.4007, 0.2000, 0.4845,
0.2667, 0.5532, 0.3333, 0.6125, 0.4000, 0.6652, 0.4667, 0.7130,
0.5333, 0.7569, 0.6000, 0.7977, 0.6667, 0.8360, 0.7333, 0.8721,
0.8000, 0.9063, 0.8667, 0.9389, 0.9333, 0.9701, 1.0000, 1.0000 ]
![sRGB tonemapping curve](android.tonemap.curveRed/srgb_tonemap.png)
</details>
<hal_details>
For good quality of mapping, at least 128 control points are
preferred.
A typical use case of this would be a gamma-1/2.2 curve, with as many
control points used as are available.
</hal_details>
</entry>
<entry name="curve" type="float" visibility="public" synthetic="true"
typedef="tonemapCurve"
hwlevel="full">
<description>Tonemapping / contrast / gamma curve to use when android.tonemap.mode
is CONTRAST_CURVE.</description>
<details>
The tonemapCurve consist of three curves for each of red, green, and blue
channels respectively. The following example uses the red channel as an
example. The same logic applies to green and blue channel.
Each channel's curve is defined by an array of control points:
curveRed =
[ P0(in, out), P1(in, out), P2(in, out), P3(in, out), ..., PN(in, out) ]
2 &lt;= N &lt;= android.tonemap.maxCurvePoints
These are sorted in order of increasing `Pin`; it is always
guaranteed that input values 0.0 and 1.0 are included in the list to
define a complete mapping. For input values between control points,
the camera device must linearly interpolate between the control
points.
Each curve can have an independent number of points, and the number
of points can be less than max (that is, the request doesn't have to
always provide a curve with number of points equivalent to
android.tonemap.maxCurvePoints).
A few examples, and their corresponding graphical mappings; these
only specify the red channel and the precision is limited to 4
digits, for conciseness.
Linear mapping:
curveRed = [ (0, 0), (1.0, 1.0) ]
![Linear mapping curve](android.tonemap.curveRed/linear_tonemap.png)
Invert mapping:
curveRed = [ (0, 1.0), (1.0, 0) ]
![Inverting mapping curve](android.tonemap.curveRed/inverse_tonemap.png)
Gamma 1/2.2 mapping, with 16 control points:
curveRed = [
(0.0000, 0.0000), (0.0667, 0.2920), (0.1333, 0.4002), (0.2000, 0.4812),
(0.2667, 0.5484), (0.3333, 0.6069), (0.4000, 0.6594), (0.4667, 0.7072),
(0.5333, 0.7515), (0.6000, 0.7928), (0.6667, 0.8317), (0.7333, 0.8685),
(0.8000, 0.9035), (0.8667, 0.9370), (0.9333, 0.9691), (1.0000, 1.0000) ]
![Gamma = 1/2.2 tonemapping curve](android.tonemap.curveRed/gamma_tonemap.png)
Standard sRGB gamma mapping, per IEC 61966-2-1:1999, with 16 control points:
curveRed = [
(0.0000, 0.0000), (0.0667, 0.2864), (0.1333, 0.4007), (0.2000, 0.4845),
(0.2667, 0.5532), (0.3333, 0.6125), (0.4000, 0.6652), (0.4667, 0.7130),
(0.5333, 0.7569), (0.6000, 0.7977), (0.6667, 0.8360), (0.7333, 0.8721),
(0.8000, 0.9063), (0.8667, 0.9389), (0.9333, 0.9701), (1.0000, 1.0000) ]
![sRGB tonemapping curve](android.tonemap.curveRed/srgb_tonemap.png)
</details>
<hal_details>
This entry is created by the framework from the curveRed, curveGreen and
curveBlue entries.
</hal_details>
</entry>
<entry name="mode" type="byte" visibility="public" enum="true"
hwlevel="full">
<enum>
<value>CONTRAST_CURVE
<notes>Use the tone mapping curve specified in
the android.tonemap.curve* entries.
All color enhancement and tonemapping must be disabled, except
for applying the tonemapping curve specified by
android.tonemap.curve.
Must not slow down frame rate relative to raw
sensor output.
</notes>
</value>
<value>FAST
<notes>
Advanced gamma mapping and color enhancement may be applied, without
reducing frame rate compared to raw sensor output.
</notes>
</value>
<value>HIGH_QUALITY
<notes>
High-quality gamma mapping and color enhancement will be applied, at
the cost of reduced frame rate compared to raw sensor output.
</notes>
</value>
</enum>
<description>High-level global contrast/gamma/tonemapping control.
</description>
<details>
When switching to an application-defined contrast curve by setting
android.tonemap.mode to CONTRAST_CURVE, the curve is defined
per-channel with a set of `(in, out)` points that specify the
mapping from input high-bit-depth pixel value to the output
low-bit-depth value. Since the actual pixel ranges of both input
and output may change depending on the camera pipeline, the values
are specified by normalized floating-point numbers.
More-complex color mapping operations such as 3D color look-up
tables, selective chroma enhancement, or other non-linear color
transforms will be disabled when android.tonemap.mode is
CONTRAST_CURVE.
This must be set to a valid mode in
android.tonemap.availableToneMapModes.
When using either FAST or HIGH_QUALITY, the camera device will
emit its own tonemap curve in android.tonemap.curve.
These values are always available, and as close as possible to the
actually used nonlinear/nonglobal transforms.
If a request is sent with CONTRAST_CURVE with the camera device's
provided curve in FAST or HIGH_QUALITY, the image's tonemap will be
roughly the same.</details>
</entry>
</controls>
<static>
<entry name="maxCurvePoints" type="int32" visibility="public"
hwlevel="full">
<description>Maximum number of supported points in the
tonemap curve that can be used for android.tonemap.curve.
</description>
<range>&amp;gt;= 64</range>
<details>
If the actual number of points provided by the application (in
android.tonemap.curve*) is less than max, the camera device will
resample the curve to its internal representation, using linear
interpolation.
The output curves in the result metadata may have a different number
of points than the input curves, and will represent the actual
hardware curves used as closely as possible when linearly interpolated.
</details>
<hal_details>
This value must be at least 64. This should be at least 128.
</hal_details>
</entry>
<entry name="availableToneMapModes" type="byte" visibility="public"
type_notes="list of enums" container="array" typedef="enumList" hwlevel="full">
<array>
<size>n</size>
</array>
<description>
The set of tonemapping modes supported by this camera device.
</description>
<details>
This tag lists the valid modes for android.tonemap.mode.
Full-capability camera devices must always support CONTRAST_CURVE and
FAST.
</details>
</entry>
</static>
<dynamic>
<clone entry="android.tonemap.curveBlue" kind="controls">
</clone>
<clone entry="android.tonemap.curveGreen" kind="controls">
</clone>
<clone entry="android.tonemap.curveRed" kind="controls">
</clone>
<clone entry="android.tonemap.curve" kind="controls">
</clone>
<clone entry="android.tonemap.mode" kind="controls">
</clone>
</dynamic>
</section>
<section name="led">
<controls>
<entry name="transmit" type="byte" visibility="hidden" optional="true"
enum="true" typedef="boolean">
<enum>
<value>OFF</value>
<value>ON</value>
</enum>
<description>This LED is nominally used to indicate to the user
that the camera is powered on and may be streaming images back to the
Application Processor. In certain rare circumstances, the OS may
disable this when video is processed locally and not transmitted to
any untrusted applications.
In particular, the LED *must* always be on when the data could be
transmitted off the device. The LED *should* always be on whenever
data is stored locally on the device.
The LED *may* be off if a trusted application is using the data that
doesn't violate the above rules.
</description>
</entry>
</controls>
<dynamic>
<clone entry="android.led.transmit" kind="controls"></clone>
</dynamic>
<static>
<entry name="availableLeds" type="byte" visibility="hidden" optional="true"
enum="true"
container="array">
<array>
<size>n</size>
</array>
<enum>
<value>TRANSMIT
<notes>android.led.transmit control is used.</notes>
</value>
</enum>
<description>A list of camera LEDs that are available on this system.
</description>
</entry>
</static>
</section>
<section name="info">
<static>
<entry name="supportedHardwareLevel" type="byte" visibility="public"
enum="true" hwlevel="legacy">
<enum>
<value>
LIMITED
<notes>
This camera device has only limited capabilities.
</notes>
</value>
<value>
FULL
<notes>
This camera device is capable of supporting advanced imaging applications.
</notes>
</value>
<value>
LEGACY
<notes>
This camera device is running in backward compatibility mode.
</notes>
</value>
</enum>
<description>
Generally classifies the overall set of the camera device functionality.
</description>
<range>Default value is LIMITED.</range>
<details>
Camera devices will come in three flavors: LEGACY, LIMITED and FULL.
A FULL device has the most support possible and will support below capabilities:
* 30fps at maximum resolution (== sensor resolution) is preferred, more than 20fps is required.
* Per frame control (android.sync.maxLatency `==` PER_FRAME_CONTROL)
* Manual sensor control (android.request.availableCapabilities contains MANUAL_SENSOR)
* Manual post-processing control (android.request.availableCapabilities contains MANUAL_POST_PROCESSING)
* Arbitrary cropping region (android.scaler.croppingType `==` FREEFORM)
* At least 3 processed (but not stalling) format output streams (android.request.maxNumOutputProc `&gt;=` 3)
* The required stream configuration defined in android.scaler.availableStreamConfigurations
* The required exposure time range defined in android.sensor.info.exposureTimeRange
* The required maxFrameDuration defined in android.sensor.info.maxFrameDuration
A LIMITED device may have some or none of the above characteristics.
To find out more refer to android.request.availableCapabilities.
A LEGACY device does not support per-frame control, manual sensor control, manual
post-processing, arbitrary cropping regions, and has relaxed performance constraints.
Each higher level supports everything the lower level supports
in this order: FULL `&gt;` LIMITED `&gt;` LEGACY.
</details>
<hal_details>
The camera 3 HAL device can implement one of two possible
operational modes; limited and full. Full support is
expected from new higher-end devices. Limited mode has
hardware requirements roughly in line with those for a
camera HAL device v1 implementation, and is expected from
older or inexpensive devices. Full is a strict superset of
limited, and they share the same essential operational flow.
For full details refer to "S3. Operational Modes" in camera3.h
Camera HAL3+ must not implement LEGACY mode. It is there
for backwards compatibility in the `android.hardware.camera2`
user-facing API only.
</hal_details>
</entry>
</static>
</section>
<section name="blackLevel">
<controls>
<entry name="lock" type="byte" visibility="public" enum="true"
typedef="boolean" hwlevel="full">
<enum>
<value>OFF</value>
<value>ON</value>
</enum>
<description> Whether black-level compensation is locked
to its current values, or is free to vary.</description>
<details>When set to ON, the values used for black-level
compensation will not change until the lock is set to
OFF.
Since changes to certain capture parameters (such as
exposure time) may require resetting of black level
compensation, the camera device must report whether setting
the black level lock was successful in the output result
metadata.
For example, if a sequence of requests is as follows:
* Request 1: Exposure = 10ms, Black level lock = OFF
* Request 2: Exposure = 10ms, Black level lock = ON
* Request 3: Exposure = 10ms, Black level lock = ON
* Request 4: Exposure = 20ms, Black level lock = ON
* Request 5: Exposure = 20ms, Black level lock = ON
* Request 6: Exposure = 20ms, Black level lock = ON
And the exposure change in Request 4 requires the camera
device to reset the black level offsets, then the output
result metadata is expected to be:
* Result 1: Exposure = 10ms, Black level lock = OFF
* Result 2: Exposure = 10ms, Black level lock = ON
* Result 3: Exposure = 10ms, Black level lock = ON
* Result 4: Exposure = 20ms, Black level lock = OFF
* Result 5: Exposure = 20ms, Black level lock = ON
* Result 6: Exposure = 20ms, Black level lock = ON
This indicates to the application that on frame 4, black
levels were reset due to exposure value changes, and pixel
values may not be consistent across captures.
The camera device will maintain the lock to the extent
possible, only overriding the lock to OFF when changes to
other request parameters require a black level recalculation
or reset.
</details>
<hal_details>
If for some reason black level locking is no longer possible
(for example, the analog gain has changed, which forces
black level offsets to be recalculated), then the HAL must
override this request (and it must report 'OFF' when this
does happen) until the next capture for which locking is
possible again.</hal_details>
<tag id="HAL2" />
</entry>
</controls>
<dynamic>
<clone entry="android.blackLevel.lock"
kind="controls">
<details>
Whether the black level offset was locked for this frame. Should be
ON if android.blackLevel.lock was ON in the capture request, unless
a change in other capture settings forced the camera device to
perform a black level reset.
</details>
</clone>
</dynamic>
</section>
<section name="sync">
<dynamic>
<entry name="frameNumber" type="int64" visibility="hidden" enum="true"
hwlevel="legacy">
<enum>
<value id="-1">CONVERGING
<notes>
The current result is not yet fully synchronized to any request.
Synchronization is in progress, and reading metadata from this
result may include a mix of data that have taken effect since the
last synchronization time.
In some future result, within android.sync.maxLatency frames,
this value will update to the actual frame number frame number
the result is guaranteed to be synchronized to (as long as the
request settings remain constant).
</notes>
</value>
<value id="-2">UNKNOWN
<notes>
The current result's synchronization status is unknown.
The result may have already converged, or it may be in
progress. Reading from this result may include some mix
of settings from past requests.
After a settings change, the new settings will eventually all
take effect for the output buffers and results. However, this
value will not change when that happens. Altering settings
rapidly may provide outcomes using mixes of settings from recent
requests.
This value is intended primarily for backwards compatibility with
the older camera implementations (for android.hardware.Camera).
</notes>
</value>
</enum>
<description>The frame number corresponding to the last request
with which the output result (metadata + buffers) has been fully
synchronized.</description>
<range>Either a non-negative value corresponding to a
`frame_number`, or one of the two enums (CONVERGING / UNKNOWN).
</range>
<details>
When a request is submitted to the camera device, there is usually a
delay of several frames before the controls get applied. A camera
device may either choose to account for this delay by implementing a
pipeline and carefully submit well-timed atomic control updates, or
it may start streaming control changes that span over several frame
boundaries.
In the latter case, whenever a request's settings change relative to
the previous submitted request, the full set of changes may take
multiple frame durations to fully take effect. Some settings may
take effect sooner (in less frame durations) than others.
While a set of control changes are being propagated, this value
will be CONVERGING.
Once it is fully known that a set of control changes have been
finished propagating, and the resulting updated control settings
have been read back by the camera device, this value will be set
to a non-negative frame number (corresponding to the request to
which the results have synchronized to).
Older camera device implementations may not have a way to detect
when all camera controls have been applied, and will always set this
value to UNKNOWN.
FULL capability devices will always have this value set to the
frame number of the request corresponding to this result.
_Further details_:
* Whenever a request differs from the last request, any future
results not yet returned may have this value set to CONVERGING (this
could include any in-progress captures not yet returned by the camera
device, for more details see pipeline considerations below).
* Submitting a series of multiple requests that differ from the
previous request (e.g. r1, r2, r3 s.t. r1 != r2 != r3)
moves the new synchronization frame to the last non-repeating
request (using the smallest frame number from the contiguous list of
repeating requests).
* Submitting the same request repeatedly will not change this value
to CONVERGING, if it was already a non-negative value.
* When this value changes to non-negative, that means that all of the
metadata controls from the request have been applied, all of the
metadata controls from the camera device have been read to the
updated values (into the result), and all of the graphics buffers
corresponding to this result are also synchronized to the request.
_Pipeline considerations_:
Submitting a request with updated controls relative to the previously
submitted requests may also invalidate the synchronization state
of all the results corresponding to currently in-flight requests.
In other words, results for this current request and up to
android.request.pipelineMaxDepth prior requests may have their
android.sync.frameNumber change to CONVERGING.
</details>
<hal_details>
Using UNKNOWN here is illegal unless android.sync.maxLatency
is also UNKNOWN.
FULL capability devices should simply set this value to the
`frame_number` of the request this result corresponds to.
</hal_details>
<tag id="V1" />
</entry>
</dynamic>
<static>
<entry name="maxLatency" type="int32" visibility="public" enum="true"
hwlevel="legacy">
<enum>
<value id="0">PER_FRAME_CONTROL
<notes>
Every frame has the requests immediately applied.
Furthermore for all results,
`android.sync.frameNumber == CaptureResult#getFrameNumber()`
Changing controls over multiple requests one after another will
produce results that have those controls applied atomically
each frame.
All FULL capability devices will have this as their maxLatency.
</notes>
</value>
<value id="-1">UNKNOWN
<notes>
Each new frame has some subset (potentially the entire set)
of the past requests applied to the camera settings.
By submitting a series of identical requests, the camera device
will eventually have the camera settings applied, but it is
unknown when that exact point will be.
All LEGACY capability devices will have this as their maxLatency.
</notes>
</value>
</enum>
<description>
The maximum number of frames that can occur after a request
(different than the previous) has been submitted, and before the
result's state becomes synchronized (by setting
android.sync.frameNumber to a non-negative value).
</description>
<units>number of processed requests</units>
<range>&amp;gt;= -1</range>
<details>
This defines the maximum distance (in number of metadata results),
between android.sync.frameNumber and the equivalent
frame number for that result.
In other words this acts as an upper boundary for how many frames
must occur before the camera device knows for a fact that the new
submitted camera settings have been applied in outgoing frames.
For example if the distance was 2,
initial request = X (repeating)
request1 = X
request2 = Y
request3 = Y
request4 = Y
where requestN has frameNumber N, and the first of the repeating
initial request's has frameNumber F (and F &lt; 1).
initial result = X' + { android.sync.frameNumber == F }
result1 = X' + { android.sync.frameNumber == F }
result2 = X' + { android.sync.frameNumber == CONVERGING }
result3 = X' + { android.sync.frameNumber == CONVERGING }
result4 = X' + { android.sync.frameNumber == 2 }
where resultN has frameNumber N.
Since `result4` has a `frameNumber == 4` and
`android.sync.frameNumber == 2`, the distance is clearly
`4 - 2 = 2`.
</details>
<hal_details>
Use `frame_count` from camera3_request_t instead of
android.request.frameCount or `CaptureResult#getFrameNumber()`.
LIMITED devices are strongly encouraged to use a non-negative
value. If UNKNOWN is used here then app developers do not have a way
to know when sensor settings have been applied.
</hal_details>
<tag id="V1" />
</entry>
</static>
</section>
</namespace>
</metadata>