Thomas G. Lane | 2cbeb8a | 1991-10-07 00:00:00 +0000 | [diff] [blame^] | 1 | The Independent JPEG Group's JPEG software |
| 2 | ========================================== |
| 3 | |
| 4 | README for release of 7-Oct-91 |
| 5 | =============================== |
| 6 | |
| 7 | This distribution contains the first public release of the Independent JPEG |
| 8 | Group's free JPEG software. You are welcome to redistribute this software and |
| 9 | to use it for any purpose, subject to the conditions under LEGAL ISSUES, below. |
| 10 | |
| 11 | This software is still undergoing revision. Updated versions may be obtained |
| 12 | by anonymous FTP to uunet.uu.net; look under directory /graphics/jpeg. This |
| 13 | particular version will be archived as jpegsrc.v1.tar.Z. If you don't have |
| 14 | access to Internet FTP, UUNET's archives are also available via UUCP; contact |
| 15 | postmaster@uunet.uu.net for information on retrieving files that way. |
| 16 | |
| 17 | Please report any problems with this software to jpeg-info@uunet.uu.net. |
| 18 | |
| 19 | If you intend to become a serious user of this software, please contact |
| 20 | jpeg-info@uunet to be added to our electronic mailing list. Then you'll be |
| 21 | notified of updates and have a chance to participate in discussions, etc. |
| 22 | |
| 23 | This software is the work of Tom Lane, Philip Gladstone, Luis Ortiz, and other |
| 24 | members of the independent JPEG group. |
| 25 | |
| 26 | |
| 27 | DISCLAIMER |
| 28 | ========== |
| 29 | |
| 30 | THIS SOFTWARE IS NOT COMPLETE NOR FULLY DEBUGGED. It is not guaranteed to be |
| 31 | useful for anything, nor to be compatible with subsequent releases, nor to be |
| 32 | an accurate implementation of the JPEG standard. (See LEGAL ISSUES for even |
| 33 | more disclaimers.) |
| 34 | |
| 35 | |
| 36 | WHAT'S HERE |
| 37 | =========== |
| 38 | |
| 39 | This distribution contains software to implement JPEG image compression and |
| 40 | decompression. JPEG is a standardized compression method for full-color and |
| 41 | gray-scale images. JPEG is intended for "real-world" scenes; cartoons and |
| 42 | other non-realistic images are not its strong suit. JPEG is lossy, meaning |
| 43 | that the output image is not necessarily identical to the input image. Hence |
| 44 | you should not use JPEG if you have to have identical output bits. However, |
| 45 | on typical images of real-world scenes, very good compression levels can be |
| 46 | obtained with hardly any visible change, and amazingly high compression levels |
| 47 | can be obtained if you can tolerate a low-quality image. For more details, |
| 48 | see the references, or just experiment with various compression settings. |
| 49 | |
| 50 | The software implements JPEG baseline and extended-sequential compression |
| 51 | processes. Provision is made for supporting all variants of these processes, |
| 52 | although some uncommon parameter settings aren't implemented yet. For legal |
| 53 | reasons, we are not distributing code for the arithmetic-coding process; see |
| 54 | LEGAL ISSUES. At present we have made no provision for supporting the |
| 55 | progressive or lossless processes defined in the standard. |
| 56 | |
| 57 | The present software is still largely in the prototype stage. It does not |
| 58 | support all possible variants of the JPEG standard, and some functions have |
| 59 | rather slow and/or crude implementations. However, it is useful already. |
| 60 | |
| 61 | The emphasis in designing this software has been on achieving portability and |
| 62 | flexibility, while also making it fast enough to be useful. We have not yet |
| 63 | undertaken serious performance measurement or tuning; we intend to do so in |
| 64 | the future. |
| 65 | |
| 66 | |
| 67 | This software can be used on several levels: |
| 68 | |
| 69 | * As canned software for JPEG compression and decompression. Just edit the |
| 70 | Makefile and configuration files as needed (see SETUP), compile and go. |
| 71 | Members of the independent JPEG group will improve the out-of-the-box |
| 72 | functionality as time goes on. |
| 73 | |
| 74 | * As the basis for other JPEG programs. For example, you could incorporate |
| 75 | the decompressor into a general image viewing package by replacing the |
| 76 | output module with write-to-screen functions. For an implementation on |
| 77 | specific hardware, you might want to replace some of the inner loops with |
| 78 | assembly code. For a non-command-line-driven system, you might want a |
| 79 | different user interface. (Members of the group will be producing Macintosh |
| 80 | and Amiga versions with appropriate user interfaces, for example.) |
| 81 | |
| 82 | * As a toolkit for experimentation with JPEG and JPEG-like algorithms. Most |
| 83 | of the individual decisions you might want to mess with are packaged up into |
| 84 | separate modules. For example, the details of color-space conversion and |
| 85 | subsampling techniques are each localized in one compressor and one |
| 86 | decompressor module. You'd probably also want to extend the user interface |
| 87 | to give you more detailed control over the JPEG compression parameters. |
| 88 | |
| 89 | In particular, we welcome the use of this software as the basis for commercial |
| 90 | products; no royalty is required. |
| 91 | |
| 92 | |
| 93 | SETUP |
| 94 | ===== |
| 95 | |
| 96 | The installation process is not very automatic; you will need at least some |
| 97 | familiarity with C programming and program build procedures for your system. |
| 98 | (Volunteers to work on improving this situation are welcome. Also, we will |
| 99 | probably start distributing pre-built binaries for popular systems at some |
| 100 | point.) |
| 101 | |
| 102 | First, select a makefile and copy it to "Makefile". "makefile.unix" |
| 103 | is appropriate for most Unix and Unix-like systems. Special makefiles are |
| 104 | included for various PC compilers. If you don't see a makefile for your |
| 105 | system, we recommend starting from makefile.unix. |
| 106 | |
| 107 | Look over the Makefile and adjust options as needed. In particular, you'll |
| 108 | need to change the CC= and CFLAGS= definitions if you don't have gcc |
| 109 | (makefile.unix only). If you have a function-prototype-less compiler, be sure |
| 110 | to uncomment the .c.o rule and say "make ansi2knr". This will cause the |
| 111 | source files to be preprocessed to change our ANSI-style function definitions |
| 112 | to old-style definitions. (Thanks to Peter Deutsch of Aladdin Enterprises for |
| 113 | ansi2knr.) |
| 114 | |
| 115 | Also look over jconfig.h and adjust #defines as necessary. If you have an |
| 116 | ANSI-compliant C compiler (gcc for instance), no changes should be necessary |
| 117 | except perhaps for RIGHT_SHIFT_IS_UNSIGNED and TWO_FILE_COMMANDLINE. For |
| 118 | older compilers other mods may be needed, depending on what ANSI features are |
| 119 | supported. If you prefer, you can usually leave jconfig.h unmodified and add |
| 120 | -D switches to the Makefile's CFLAGS= definition. |
| 121 | |
| 122 | Then say "make". |
| 123 | |
| 124 | If you have trouble with missing system include files or inclusion of the |
| 125 | wrong ones, you can fix it in jinclude.h. In particular, if you are using |
| 126 | gcc on a machine with non-ANSI system include files, you are likely to find |
| 127 | that jinclude.h tries to include the wrong files (because gcc defines |
| 128 | __STDC__). There's no good automatic solution to this, so you'll just have |
| 129 | to hand-edit jinclude.h. |
| 130 | |
| 131 | As a quick test of functionality we've included three sample files: |
| 132 | testorig.jpg same as blkint.jpg from JPEG validation floppy. |
| 133 | testimg.ppm output of djpeg testorig.jpg |
| 134 | testimg.jpg output of cjpeg testimg.ppm |
| 135 | The two .jpg files aren't identical due to different parameter choices (and |
| 136 | wouldn't be anyway, since JPEG is lossy). However, if you can generate |
| 137 | duplicates of testimg.ppm and testimg.jpg then you probably have a working |
| 138 | port. "make test" will perform the necessary comparisons (by generating |
| 139 | testout.ppm and testout.jpg and comparing these to testimg.*). NOTE: this |
| 140 | is far from an exhaustive test of the JPEG software; some modules, such as |
| 141 | color quantization and GIF I/O, are not exercised at all. It's just a quick |
| 142 | test to give you some confidence that you haven't missed something major. |
| 143 | |
| 144 | If you need to make a smaller version of the JPEG software, some optional |
| 145 | functions can be removed at compile time. See the xxx_SUPPORTED #defines |
| 146 | in jconfig.h. (Not a lot is actually removed right now, but as more optional |
| 147 | stuff gets added, this mechanism will start to make a difference.) |
| 148 | |
| 149 | If you want to incorporate the JPEG code as subroutines in a larger program, |
| 150 | we recommend that you make libjpeg.a. Then use the .h files and libjpeg.a as |
| 151 | your interface to the JPEG functions. Your surrounding program will have to |
| 152 | provide functionality similar to what's in jcmain.c or jdmain.c, and you may |
| 153 | want to replace jerror.c and possibly other modules depending on your needs. |
| 154 | See the "architecture" file for more info. If it seems to you that the system |
| 155 | structure doesn't accommodate what you want to do, please contact the authors. |
| 156 | |
| 157 | Special notes for Macintosh Think C users: If you have version 5.0 you should |
| 158 | be able to just turn on __STDC__ through the compiler switch that enables |
| 159 | that. With version 4.0 you must manually edit jconfig.h to define PROTO, |
| 160 | HAVE_UNSIGNED_CHAR, HAVE_UNSIGNED_SHORT, and const. (It seems to be safe to |
| 161 | just define __STDC__ to take care of the first three.) When setting up |
| 162 | project files, use the COBJECTS and DOBJECTS lists in makefile.unix as a guide |
| 163 | to which files need to be included, and add the ANSI and Unix C libraries in a |
| 164 | separate segment. You may need to divide the JPEG files into more than one |
| 165 | segment; you can do this pretty much as you please. |
| 166 | |
| 167 | |
| 168 | USAGE |
| 169 | ===== |
| 170 | |
| 171 | The user interface is pretty minimal at this point. We haven't bothered to |
| 172 | generate manual-page files since the switches badly need redesign. At the |
| 173 | moment, things work like this: |
| 174 | |
| 175 | There are two programs, cjpeg to compress an image file into JPEG format, |
| 176 | and djpeg to decompress. |
| 177 | |
| 178 | On Unix systems, you say: |
| 179 | cjpeg [switches] [imagefile] >jpegfile |
| 180 | djpeg [switches] [jpegfile] >imagefile |
| 181 | The programs read the specified input file, or standard input if none is |
| 182 | named. They always write to standard output (with trace/error messages to |
| 183 | standard error). These conventions are handy for piping images between |
| 184 | programs. |
| 185 | |
| 186 | On PC, Macintosh, and Amiga systems, you say: |
| 187 | cjpeg [switches] imagefile jpegfile |
| 188 | djpeg [switches] jpegfile imagefile |
| 189 | i.e., both input and output files are named on the command line. This style |
| 190 | is a little more foolproof, and it loses no functionality if you don't have |
| 191 | pipes. You can get this style on Unix too, if you prefer, by defining |
| 192 | TWO_FILE_COMMANDLINE in jconfig.h or in the Makefile. You MUST use this style |
| 193 | on any system that doesn't cope well with binary data fed through |
| 194 | stdin/stdout. |
| 195 | |
| 196 | Currently supported image file formats include raw-format PPM, raw-format PGM |
| 197 | (for monochrome images), and GIF. cjpeg recognizes the input image format |
| 198 | automatically, but you have to tell djpeg which format to generate. |
| 199 | |
| 200 | The only JPEG file format currently supported is a raw JPEG data stream. |
| 201 | Unless modified, the programs use the JFIF conventions for variables left |
| 202 | unspecified by the JPEG standard. (In particular, cjpeg generates a JFIF APP0 |
| 203 | marker.) Support for the JPEG-in-TIFF format will probably be added at some |
| 204 | future date. |
| 205 | |
| 206 | The command line switches for cjpeg are: |
| 207 | |
| 208 | -I Generate noninterleaved JPEG file (not yet supported). |
| 209 | |
| 210 | -Q quality Scale quantization tables to adjust quality. |
| 211 | Quality is 0 (worst) to 100 (best); default is 75. |
| 212 | (See below for more info.) |
| 213 | |
| 214 | -a Use arithmetic coding rather than Huffman coding. |
| 215 | (Not currently supported, see LEGAL ISSUES.) |
| 216 | |
| 217 | -o Perform optimization of entropy encoding parameters. |
| 218 | Without this, default Huffman or arithmetic |
| 219 | parameters are used. -o makes the JPEG file a tad |
| 220 | smaller, but compression uses much more memory. |
| 221 | Image quality is unaffected by -o. |
| 222 | |
| 223 | -d Enable debug printout. More -d's give more printout. |
| 224 | |
| 225 | Typically you'd use -Q settings of 50 or 75 or so. -Q 100 will generate a |
| 226 | quantization table of all 1's, meaning no quantization loss; then any |
| 227 | differences between input and output images are due to subsampling or to |
| 228 | roundoff error in the DCT or colorspace-conversion steps. -Q values below 50 |
| 229 | may be useful for making real small, low-quality images. Try -Q 2 (or so) for |
| 230 | some amusing Cubist effects. (Note that -Q values below about 25 generate |
| 231 | 2-byte quantization tables, which are not decodable by pure baseline JPEG |
| 232 | decoders. cjpeg emits a warning message when you give such a -Q value.) |
| 233 | |
| 234 | The command line switches for djpeg are: |
| 235 | |
| 236 | -G Select GIF output format (implies -q, with default |
| 237 | of 256 colors). |
| 238 | |
| 239 | -b Perform cross-block smoothing. This is quite |
| 240 | memory-intensive and only seems to improve the image |
| 241 | at very low quality settings (-Q 10 to 20 or so). |
| 242 | |
| 243 | -g Force gray-scale output even if input is color. |
| 244 | |
| 245 | -q N Quantize to N colors. |
| 246 | |
| 247 | -D Use Floyd-Steinberg dithering in color quantization. |
| 248 | |
| 249 | -2 Use two-pass color quantization (not yet supported). |
| 250 | |
| 251 | -d Enable debug printout. More -d's give more printout. |
| 252 | |
| 253 | Color quantization currently uses a rather shoddy algorithm (although it's not |
| 254 | so horrible when dithered). Because of this, the GIF output mode is not |
| 255 | recommended in the current release, except for gray-scale output. You can get |
| 256 | better results by applying ppmquant to the unquantized (PPM) output of djpeg, |
| 257 | then converting to GIF with ppmtogif. We expect to provide a considerably |
| 258 | better quantization algorithm in a future release. |
| 259 | |
| 260 | Note that djpeg *can* read noninterleaved JPEG files even though cjpeg can't |
| 261 | yet generate them. For most applications this is a nonissue, since hardly |
| 262 | anybody seems to be using noninterleaved format. |
| 263 | |
| 264 | On a non-virtual-memory machine, you may run out of memory if you use -I or -o |
| 265 | in cjpeg, or -q ... -2 in djpeg, or try to read an interlaced GIF file. This |
| 266 | will be addressed eventually by replacing jvirtmem.c with something that uses |
| 267 | temporary files for large images (see TO DO). |
| 268 | |
| 269 | |
| 270 | REFERENCES |
| 271 | ========== |
| 272 | |
| 273 | The best and most readily available introduction to the JPEG compression |
| 274 | algorithm is Wallace's article in the April '91 CACM: |
| 275 | Wallace, Gregory K. "The JPEG Still Picture Compression Standard", |
| 276 | Communications of the ACM, April 1991 (vol. 34 no. 4), pp. 30-44. |
| 277 | (Adjacent articles in that issue discuss MPEG motion picture compression, |
| 278 | applications of JPEG, and related topics.) We highly recommend reading that |
| 279 | article before looking at any of the JPEG software. |
| 280 | |
| 281 | For more detail about the JPEG standard you pretty much have to go to the |
| 282 | draft standard, which is not nearly as intelligible as Wallace's article. |
| 283 | The current version is ISO/IEC Committee Draft CD 10918-1 dated 1991-03-15. |
| 284 | The standard is not presently available electronically; you must order a paper |
| 285 | copy through ISO. |
| 286 | |
| 287 | The JPEG standard does not specify all details of an interchangeable file |
| 288 | format. For the omitted details we follow the "JFIF" conventions, revision |
| 289 | 1.01. A copy of the JFIF spec is available from: |
| 290 | Literature Department |
| 291 | C-Cube Microsystems, Inc. |
| 292 | 399A West Trimble Road |
| 293 | San Jose, CA 95131 |
| 294 | (408) 944-6300 |
| 295 | Requests can also be e-mailed to info@c3.pla.ca.us (this address good after |
| 296 | 10/10/91). The same source can supply copies of the draft JPEG-in-TIFF specs. |
| 297 | |
| 298 | If you want to understand this implementation, start by reading the |
| 299 | "architecture" documentation file. Please read "codingrules" if you want to |
| 300 | contribute any code. |
| 301 | |
| 302 | |
| 303 | SUPPORTING SOFTWARE |
| 304 | =================== |
| 305 | |
| 306 | You will probably want Jef Poskanzer's PBMPLUS image software; this provides |
| 307 | many useful operations on PPM-format image files. In particular, it can |
| 308 | convert PPM images to and from a wide range of other formats. You can FTP |
| 309 | this free software from export.lcs.mit.edu (contrib/pbmplus*.tar.Z) or |
| 310 | ftp.ee.lbl.gov (pbmplus*.tar.Z). |
| 311 | |
| 312 | If you are using X Windows you might want to use the xv or xloadimage viewers |
| 313 | to save yourself the trouble of converting PPM to some other format. |
| 314 | Both of these can be found in the contrib directory at export.lcs.mit.edu. |
| 315 | |
| 316 | |
| 317 | LEGAL ISSUES |
| 318 | ============ |
| 319 | |
| 320 | The authors make NO WARRANTY or representation, either express or implied, |
| 321 | with respect to this software, its quality, accuracy, merchantability, or |
| 322 | fitness for a particular purpose. This software is provided "AS IS", and you, |
| 323 | its user, assume the entire risk as to its quality and accuracy. |
| 324 | |
| 325 | This software is copyright (C) 1991, Thomas G. Lane. |
| 326 | All Rights Reserved except as specified below. |
| 327 | |
| 328 | Permission is hereby granted to use, copy, modify, and distribute this |
| 329 | software (or portions thereof) for any purpose, without fee, subject to these |
| 330 | conditions: |
| 331 | (1) If any part of the source code for this software is distributed, then this |
| 332 | README file must be included, with this copyright and no-warranty notice |
| 333 | unaltered; and any additions, deletions, or changes to the original files |
| 334 | must be clearly indicated in accompanying documentation. |
| 335 | (2) If only executable code is distributed, then the accompanying |
| 336 | documentation must state that "this software is based in part on the work of |
| 337 | the Independent JPEG Group". |
| 338 | (3) Permission for use of this software is granted only if the user accepts |
| 339 | full responsibility for any undesirable consequences; the authors accept |
| 340 | NO LIABILITY for damages of any kind. |
| 341 | |
| 342 | Permission is NOT granted for the use of any author's name or author's company |
| 343 | name in advertising or publicity relating to this software or products derived |
| 344 | from it. This software may be referred to only as "the Independent JPEG |
| 345 | Group's software". |
| 346 | |
| 347 | We specifically permit and encourage the use of this software as the basis of |
| 348 | commercial products, provided that all warranty or liability claims are |
| 349 | assumed by the product vendor. |
| 350 | |
| 351 | |
| 352 | ansi2knr.c is included in this distribution by permission of L. Peter Deutsch, |
| 353 | sole proprietor of its copyright holder, Aladdin Enterprises of Menlo Park, CA. |
| 354 | ansi2knr.c is NOT covered by the above copyright and conditions, but instead |
| 355 | by the usual distribution terms of the Free Software Foundation; principally, |
| 356 | that you must include source code if you redistribute it. (See the file |
| 357 | ansi2knr.c for full details.) However, since ansi2knr.c is not needed as part |
| 358 | of any product generated from the JPEG code, this does not limit you more than |
| 359 | the foregoing paragraphs do. |
| 360 | |
| 361 | |
| 362 | It appears that the arithmetic coding option of the JPEG spec is covered by |
| 363 | patents held by IBM, and possibly also patents of AT&T and Mitsubishi. Hence |
| 364 | arithmetic coding cannot legally be used without obtaining one or more |
| 365 | licenses. For this reason, support for arithmetic coding has been removed |
| 366 | from the free JPEG software. (Since arithmetic coding provides only a |
| 367 | marginal gain over the unpatented Huffman mode, it is unlikely that very many |
| 368 | people will choose to use it. If you do obtain such a license, contact |
| 369 | jpeg-info@uunet.uu.net for a copy of our arithmetic coding modules.) So far |
| 370 | as we are aware, there are no patent restrictions on the remaining code. |
| 371 | |
| 372 | |
| 373 | TO DO |
| 374 | ===== |
| 375 | |
| 376 | Many of the modules need fleshing out to provide more complete |
| 377 | implementations, or to provide faster paths for common cases. The greatest |
| 378 | needs are for (a) decent color quantization, and (b) a memory manager |
| 379 | implementation that can work in limited memory by swapping "big" images to |
| 380 | temporary files. I (Tom Lane) am going to work on color quantization next. |
| 381 | Volunteers to write a PC memory manager, or to work on any other modules, are |
| 382 | welcome. |
| 383 | |
| 384 | We'd appreciate it if people would compile and check out the code on as wide a |
| 385 | variety of systems as possible, and report any portability problems |
| 386 | encountered (with solutions, if possible). Checks of file compatibility with |
| 387 | other JPEG implementations would also be of interest. Finally, we would |
| 388 | appreciate code profiles showing where the most time is spent, especially on |
| 389 | unusual systems. |
| 390 | |
| 391 | Please send bug reports, offers of help, etc. to jpeg-info@uunet.uu.net. |