1 NOTE: This file was modified by The libjpeg-turbo Project to include only
2 information relevant to libjpeg-turbo and to wordsmith certain sections.
4 USAGE instructions for the Independent JPEG Group's JPEG software
5 =================================================================
7 This file describes usage of the JPEG conversion programs cjpeg and djpeg,
8 as well as the utility programs jpegtran, rdjpgcom and wrjpgcom. (See
9 the other documentation files if you wish to use the JPEG library within
12 If you are on a Unix machine you may prefer to read the Unix-style manual
13 pages in files cjpeg.1, djpeg.1, jpegtran.1, rdjpgcom.1, wrjpgcom.1.
18 These programs implement JPEG image encoding, decoding, and transcoding.
19 JPEG (pronounced "jay-peg") is a standardized compression method for
20 full-color and grayscale images.
25 We provide two programs, cjpeg to compress an image file into JPEG format,
26 and djpeg to decompress a JPEG file back into a conventional image format.
28 On most systems, you say:
29 cjpeg [switches] [imagefile] >jpegfile
31 djpeg [switches] [jpegfile] >imagefile
32 The programs read the specified input file, or standard input if none is
33 named. They always write to standard output (with trace/error messages to
34 standard error). These conventions are handy for piping images between
37 If you defined TWO_FILE_COMMANDLINE when compiling the programs, you can
39 cjpeg [switches] imagefile jpegfile
41 djpeg [switches] jpegfile imagefile
42 i.e., both the input and output files are named on the command line. This
43 style is a little more foolproof, and it loses no functionality if you don't
47 cjpeg [switches] -outfile jpegfile imagefile
49 djpeg [switches] -outfile imagefile jpegfile
50 This syntax works on all systems, so it is useful for scripts.
52 The currently supported image file formats are: PPM (PBMPLUS color format),
53 PGM (PBMPLUS grayscale format), BMP, GIF, and Targa. cjpeg recognizes the
54 input image format automatically, with the exception of some Targa files. You
55 have to tell djpeg which format to generate.
57 JPEG files are in the defacto standard JFIF file format. There are other,
58 less widely used JPEG-based file formats, but we don't support them.
60 All switch names may be abbreviated; for example, -grayscale may be written
61 -gray or -gr. Most of the "basic" switches can be abbreviated to as little as
62 one letter. Upper and lower case are equivalent (-BMP is the same as -bmp).
63 British spellings are also accepted (e.g., -greyscale), though for brevity
64 these are not mentioned below.
69 The basic command line switches for cjpeg are:
71 -quality N[,...] Scale quantization tables to adjust image quality.
72 Quality is 0 (worst) to 100 (best); default is 75.
73 (See below for more info.)
75 -grayscale Create monochrome JPEG file from color input. By
76 saying -grayscale, you'll get a smaller JPEG file that
77 takes less time to process.
79 -rgb Create RGB JPEG file.
80 Using this switch suppresses the conversion from RGB
81 colorspace input to the default YCbCr JPEG colorspace.
83 -optimize Perform optimization of entropy encoding parameters.
84 Without this, default encoding parameters are used.
85 -optimize usually makes the JPEG file a little smaller,
86 but cjpeg runs somewhat slower and needs much more
87 memory. Image quality and speed of decompression are
88 unaffected by -optimize.
90 -progressive Create progressive JPEG file (see below).
92 -targa Input file is Targa format. Targa files that contain
93 an "identification" field will not be automatically
94 recognized by cjpeg; for such files you must specify
95 -targa to make cjpeg treat the input as Targa format.
96 For most Targa files, you won't need this switch.
98 The -quality switch lets you trade off compressed file size against quality of
99 the reconstructed image: the higher the quality setting, the larger the JPEG
100 file, and the closer the output image will be to the original input. Normally
101 you want to use the lowest quality setting (smallest file) that decompresses
102 into something visually indistinguishable from the original image. For this
103 purpose the quality setting should generally be between 50 and 95 (the default
104 is 75) for photographic images. If you see defects at -quality 75, then go up
105 5 or 10 counts at a time until you are happy with the output image. (The
106 optimal setting will vary from one image to another.)
108 -quality 100 will generate a quantization table of all 1's, minimizing loss
109 in the quantization step (but there is still information loss in subsampling,
110 as well as roundoff error.) For most images, specifying a quality value above
111 about 95 will increase the size of the compressed file dramatically, and while
112 the quality gain from these higher quality values is measurable (using metrics
113 such as PSNR or SSIM), it is rarely perceivable by human vision.
115 In the other direction, quality values below 50 will produce very small files
116 of low image quality. Settings around 5 to 10 might be useful in preparing an
117 index of a large image library, for example. Try -quality 2 (or so) for some
118 amusing Cubist effects. (Note: quality values below about 25 generate 2-byte
119 quantization tables, which are considered optional in the JPEG standard.
120 cjpeg emits a warning message when you give such a quality value, because some
121 other JPEG programs may be unable to decode the resulting file. Use -baseline
122 if you need to ensure compatibility at low quality values.)
124 The -quality option has been extended in this version of cjpeg to support
125 separate quality settings for luminance and chrominance (or, in general,
126 separate settings for every quantization table slot.) The principle is the
127 same as chrominance subsampling: since the human eye is more sensitive to
128 spatial changes in brightness than spatial changes in color, the chrominance
129 components can be quantized more than the luminance components without
130 incurring any visible image quality loss. However, unlike subsampling, this
131 feature reduces data in the frequency domain instead of the spatial domain,
132 which allows for more fine-grained control. This option is useful in
133 quality-sensitive applications, for which the artifacts generated by
134 subsampling may be unacceptable.
136 The -quality option accepts a comma-separated list of parameters, which
137 respectively refer to the quality levels that should be assigned to the
138 quantization table slots. If there are more q-table slots than parameters,
139 then the last parameter is replicated. Thus, if only one quality parameter is
140 given, this is used for both luminance and chrominance (slots 0 and 1,
141 respectively), preserving the legacy behavior of cjpeg v6b and prior. More (or
142 customized) quantization tables can be set with the -qtables option and
143 assigned to components with the -qslots option (see the "wizard" switches
146 JPEG files generated with separate luminance and chrominance quality are fully
147 compliant with standard JPEG decoders.
149 CAUTION: For this setting to be useful, be sure to pass an argument of
150 -sample 1x1 to cjpeg to disable chrominance subsampling. Otherwise, the
151 default subsampling level (2x2, AKA "4:2:0") will be used.
153 The -progressive switch creates a "progressive JPEG" file. In this type of
154 JPEG file, the data is stored in multiple scans of increasing quality. If the
155 file is being transmitted over a slow communications link, the decoder can use
156 the first scan to display a low-quality image very quickly, and can then
157 improve the display with each subsequent scan. The final image is exactly
158 equivalent to a standard JPEG file of the same quality setting, and the total
159 file size is about the same --- often a little smaller.
161 Switches for advanced users:
163 -arithmetic Use arithmetic coding. CAUTION: arithmetic coded JPEG
164 is not yet widely implemented, so many decoders will
165 be unable to view an arithmetic coded JPEG file at
168 -dct int Use accurate integer DCT method (default).
169 -dct fast Use less accurate integer DCT method [legacy feature].
170 When the Independent JPEG Group's software was first
171 released in 1991, the compression time for a
172 1-megapixel JPEG image on a mainstream PC was measured
173 in minutes. Thus, the fast integer DCT algorithm
174 provided noticeable performance benefits. On modern
175 CPUs running libjpeg-turbo, however, the compression
176 time for a 1-megapixel JPEG image is measured in
177 milliseconds, and thus the performance benefits of the
178 fast algorithm are much less noticeable. On modern
179 x86/x86-64 CPUs that support AVX2 instructions, the
180 fast and int methods have similar performance. On
181 other types of CPUs, the fast method is generally about
182 5-15% faster than the int method.
184 For quality levels of 90 and below, there should be
185 little or no perceptible quality difference between the
186 two algorithms. For quality levels above 90, however,
187 the difference between the fast and int methods becomes
188 more pronounced. With quality=97, for instance, the
189 fast method incurs generally about a 1-3 dB loss in
190 PSNR relative to the int method, but this can be larger
191 for some images. Do not use the fast method with
192 quality levels above 97. The algorithm often
193 degenerates at quality=98 and above and can actually
194 produce a more lossy image than if lower quality levels
195 had been used. Also, in libjpeg-turbo, the fast method
196 is not fully accelerated for quality levels above 97,
197 so it will be slower than the int method.
198 -dct float Use floating-point DCT method [legacy feature].
199 The float method does not produce significantly more
200 accurate results than the int method, and it is much
201 slower. The float method may also give different
202 results on different machines due to varying roundoff
203 behavior, whereas the integer methods should give the
204 same results on all machines.
206 -restart N Emit a JPEG restart marker every N MCU rows, or every
207 N MCU blocks if "B" is attached to the number.
208 -restart 0 (the default) means no restart markers.
210 -smooth N Smooth the input image to eliminate dithering noise.
211 N, ranging from 1 to 100, indicates the strength of
212 smoothing. 0 (the default) means no smoothing.
214 -maxmemory N Set limit for amount of memory to use in processing
215 large images. Value is in thousands of bytes, or
216 millions of bytes if "M" is attached to the number.
217 For example, -max 4m selects 4000000 bytes. If more
218 space is needed, an error will occur.
220 -verbose Enable debug printout. More -v's give more printout.
221 or -debug Also, version information is printed at startup.
223 The -restart option inserts extra markers that allow a JPEG decoder to
224 resynchronize after a transmission error. Without restart markers, any damage
225 to a compressed file will usually ruin the image from the point of the error
226 to the end of the image; with restart markers, the damage is usually confined
227 to the portion of the image up to the next restart marker. Of course, the
228 restart markers occupy extra space. We recommend -restart 1 for images that
229 will be transmitted across unreliable networks such as Usenet.
231 The -smooth option filters the input to eliminate fine-scale noise. This is
232 often useful when converting dithered images to JPEG: a moderate smoothing
233 factor of 10 to 50 gets rid of dithering patterns in the input file, resulting
234 in a smaller JPEG file and a better-looking image. Too large a smoothing
235 factor will visibly blur the image, however.
237 Switches for wizards:
239 -baseline Force baseline-compatible quantization tables to be
240 generated. This clamps quantization values to 8 bits
241 even at low quality settings. (This switch is poorly
242 named, since it does not ensure that the output is
243 actually baseline JPEG. For example, you can use
244 -baseline and -progressive together.)
246 -qtables file Use the quantization tables given in the specified
249 -qslots N[,...] Select which quantization table to use for each color
252 -sample HxV[,...] Set JPEG sampling factors for each color component.
254 -scans file Use the scan script given in the specified text file.
256 The "wizard" switches are intended for experimentation with JPEG. If you
257 don't know what you are doing, DON'T USE THEM. These switches are documented
258 further in the file wizard.txt.
263 The basic command line switches for djpeg are:
265 -colors N Reduce image to at most N colors. This reduces the
266 or -quantize N number of colors used in the output image, so that it
267 can be displayed on a colormapped display or stored in
268 a colormapped file format. For example, if you have
269 an 8-bit display, you'd need to reduce to 256 or fewer
270 colors. (-colors is the recommended name, -quantize
271 is provided only for backwards compatibility.)
273 -fast Select recommended processing options for fast, low
274 quality output. (The default options are chosen for
275 highest quality output.) Currently, this is equivalent
276 to "-dct fast -nosmooth -onepass -dither ordered".
278 -grayscale Force grayscale output even if JPEG file is color.
279 Useful for viewing on monochrome displays; also,
280 djpeg runs noticeably faster in this mode.
282 -rgb Force RGB output even if JPEG file is grayscale.
284 -scale M/N Scale the output image by a factor M/N. Currently
285 the scale factor must be M/8, where M is an integer
286 between 1 and 16 inclusive, or any reduced fraction
287 thereof (such as 1/2, 3/4, etc. Scaling is handy if
288 the image is larger than your screen; also, djpeg runs
289 much faster when scaling down the output.
291 -bmp Select BMP output format (Windows flavor). 8-bit
292 colormapped format is emitted if -colors or -grayscale
293 is specified, or if the JPEG file is grayscale;
294 otherwise, 24-bit full-color format is emitted.
296 -gif Select GIF output format (LZW-compressed). Since GIF
297 does not support more than 256 colors, -colors 256 is
298 assumed (unless you specify a smaller number of
299 colors). If you specify -fast, the default number of
302 -gif0 Select GIF output format (uncompressed). Since GIF
303 does not support more than 256 colors, -colors 256 is
304 assumed (unless you specify a smaller number of
305 colors). If you specify -fast, the default number of
308 -os2 Select BMP output format (OS/2 1.x flavor). 8-bit
309 colormapped format is emitted if -colors or -grayscale
310 is specified, or if the JPEG file is grayscale;
311 otherwise, 24-bit full-color format is emitted.
313 -pnm Select PBMPLUS (PPM/PGM) output format (this is the
314 default format). PGM is emitted if the JPEG file is
315 grayscale or if -grayscale is specified; otherwise
318 -targa Select Targa output format. Grayscale format is
319 emitted if the JPEG file is grayscale or if
320 -grayscale is specified; otherwise, colormapped format
321 is emitted if -colors is specified; otherwise, 24-bit
322 full-color format is emitted.
324 Switches for advanced users:
326 -dct int Use accurate integer DCT method (default).
327 -dct fast Use less accurate integer DCT method [legacy feature].
328 When the Independent JPEG Group's software was first
329 released in 1991, the decompression time for a
330 1-megapixel JPEG image on a mainstream PC was measured
331 in minutes. Thus, the fast integer DCT algorithm
332 provided noticeable performance benefits. On modern
333 CPUs running libjpeg-turbo, however, the decompression
334 time for a 1-megapixel JPEG image is measured in
335 milliseconds, and thus the performance benefits of the
336 fast algorithm are much less noticeable. On modern
337 x86/x86-64 CPUs that support AVX2 instructions, the
338 fast and int methods have similar performance. On
339 other types of CPUs, the fast method is generally about
340 5-15% faster than the int method.
342 If the JPEG image was compressed using a quality level
343 of 85 or below, then there should be little or no
344 perceptible quality difference between the two
345 algorithms. When decompressing images that were
346 compressed using quality levels above 85, however, the
347 difference between the fast and int methods becomes
348 more pronounced. With images compressed using
349 quality=97, for instance, the fast method incurs
350 generally about a 4-6 dB loss in PSNR relative to the
351 int method, but this can be larger for some images. If
352 you can avoid it, do not use the fast method when
353 decompressing images that were compressed using quality
354 levels above 97. The algorithm often degenerates for
355 such images and can actually produce a more lossy
356 output image than if the JPEG image had been compressed
357 using lower quality levels.
358 -dct float Use floating-point DCT method [legacy feature].
359 The float method does not produce significantly more
360 accurate results than the int method, and it is much
361 slower. The float method may also give different
362 results on different machines due to varying roundoff
363 behavior, whereas the integer methods should give the
364 same results on all machines.
366 -dither fs Use Floyd-Steinberg dithering in color quantization.
367 -dither ordered Use ordered dithering in color quantization.
368 -dither none Do not use dithering in color quantization.
369 By default, Floyd-Steinberg dithering is applied when
370 quantizing colors; this is slow but usually produces
371 the best results. Ordered dither is a compromise
372 between speed and quality; no dithering is fast but
373 usually looks awful. Note that these switches have
374 no effect unless color quantization is being done.
375 Ordered dither is only available in -onepass mode.
377 -map FILE Quantize to the colors used in the specified image
378 file. This is useful for producing multiple files
379 with identical color maps, or for forcing a predefined
380 set of colors to be used. The FILE must be a GIF
381 or PPM file. This option overrides -colors and
384 -nosmooth Use a faster, lower-quality upsampling routine.
386 -onepass Use one-pass instead of two-pass color quantization.
387 The one-pass method is faster and needs less memory,
388 but it produces a lower-quality image. -onepass is
389 ignored unless you also say -colors N. Also,
390 the one-pass method is always used for grayscale
391 output (the two-pass method is no improvement then).
393 -maxmemory N Set limit for amount of memory to use in processing
394 large images. Value is in thousands of bytes, or
395 millions of bytes if "M" is attached to the number.
396 For example, -max 4m selects 4000000 bytes. If more
397 space is needed, an error will occur.
399 -verbose Enable debug printout. More -v's give more printout.
400 or -debug Also, version information is printed at startup.
405 Color GIF files are not the ideal input for JPEG; JPEG is really intended for
406 compressing full-color (24-bit) images. In particular, don't try to convert
407 cartoons, line drawings, and other images that have only a few distinct
408 colors. GIF works great on these, JPEG does not. If you want to convert a
409 GIF to JPEG, you should experiment with cjpeg's -quality and -smooth options
410 to get a satisfactory conversion. -smooth 10 or so is often helpful.
412 Avoid running an image through a series of JPEG compression/decompression
413 cycles. Image quality loss will accumulate; after ten or so cycles the image
414 may be noticeably worse than it was after one cycle. It's best to use a
415 lossless format while manipulating an image, then convert to JPEG format when
416 you are ready to file the image away.
418 The -optimize option to cjpeg is worth using when you are making a "final"
419 version for posting or archiving. It's also a win when you are using low
420 quality settings to make very small JPEG files; the percentage improvement
421 is often a lot more than it is on larger files. (At present, -optimize
422 mode is always selected when generating progressive JPEG files.)
427 To get a quick preview of an image, use the -grayscale and/or -scale switches.
428 "-grayscale -scale 1/8" is the fastest case.
430 Several options are available that trade off image quality to gain speed.
431 "-fast" turns on the recommended settings.
433 "-dct fast" and/or "-nosmooth" gain speed at a small sacrifice in quality.
434 When producing a color-quantized image, "-onepass -dither ordered" is fast but
435 much lower quality than the default behavior. "-dither none" may give
436 acceptable results in two-pass mode, but is seldom tolerable in one-pass mode.
439 HINTS FOR BOTH PROGRAMS
441 If the memory needed by cjpeg or djpeg exceeds the limit specified by
442 -maxmemory, an error will occur. You can leave out -progressive and -optimize
443 (for cjpeg) or specify -onepass (for djpeg) to reduce memory usage.
445 On machines that have "environment" variables, you can define the environment
446 variable JPEGMEM to set the default memory limit. The value is specified as
447 described for the -maxmemory switch. JPEGMEM overrides the default value
448 specified when the program was compiled, and itself is overridden by an
449 explicit -maxmemory switch.
454 jpegtran performs various useful transformations of JPEG files.
455 It can translate the coded representation from one variant of JPEG to another,
456 for example from baseline JPEG to progressive JPEG or vice versa. It can also
457 perform some rearrangements of the image data, for example turning an image
458 from landscape to portrait format by rotation. For EXIF files and JPEG files
459 containing Exif data, you may prefer to use exiftran instead.
461 jpegtran works by rearranging the compressed data (DCT coefficients), without
462 ever fully decoding the image. Therefore, its transformations are lossless:
463 there is no image degradation at all, which would not be true if you used
464 djpeg followed by cjpeg to accomplish the same conversion. But by the same
465 token, jpegtran cannot perform lossy operations such as changing the image
466 quality. However, while the image data is losslessly transformed, metadata
467 can be removed. See the -copy option for specifics.
469 jpegtran uses a command line syntax similar to cjpeg or djpeg.
470 On most systems, you say:
471 jpegtran [switches] [inputfile] >outputfile
472 If you defined TWO_FILE_COMMANDLINE when compiling the program, you can instead
474 jpegtran [switches] inputfile outputfile
475 where both the input and output files are JPEG files.
477 To specify the coded JPEG representation used in the output file,
478 jpegtran accepts a subset of the switches recognized by cjpeg:
479 -optimize Perform optimization of entropy encoding parameters.
480 -progressive Create progressive JPEG file.
481 -arithmetic Use arithmetic coding.
482 -restart N Emit a JPEG restart marker every N MCU rows, or every
483 N MCU blocks if "B" is attached to the number.
484 -scans file Use the scan script given in the specified text file.
485 See the previous discussion of cjpeg for more details about these switches.
486 If you specify none of these switches, you get a plain baseline-JPEG output
487 file. The quality setting and so forth are determined by the input file.
489 The image can be losslessly transformed by giving one of these switches:
490 -flip horizontal Mirror image horizontally (left-right).
491 -flip vertical Mirror image vertically (top-bottom).
492 -rotate 90 Rotate image 90 degrees clockwise.
493 -rotate 180 Rotate image 180 degrees.
494 -rotate 270 Rotate image 270 degrees clockwise (or 90 ccw).
495 -transpose Transpose image (across UL-to-LR axis).
496 -transverse Transverse transpose (across UR-to-LL axis).
498 The transpose transformation has no restrictions regarding image dimensions.
499 The other transformations operate rather oddly if the image dimensions are not
500 a multiple of the iMCU size (usually 8 or 16 pixels), because they can only
501 transform complete blocks of DCT coefficient data in the desired way.
503 jpegtran's default behavior when transforming an odd-size image is designed
504 to preserve exact reversibility and mathematical consistency of the
505 transformation set. As stated, transpose is able to flip the entire image
506 area. Horizontal mirroring leaves any partial iMCU column at the right edge
507 untouched, but is able to flip all rows of the image. Similarly, vertical
508 mirroring leaves any partial iMCU row at the bottom edge untouched, but is
509 able to flip all columns. The other transforms can be built up as sequences
510 of transpose and flip operations; for consistency, their actions on edge
511 pixels are defined to be the same as the end result of the corresponding
512 transpose-and-flip sequence.
514 For practical use, you may prefer to discard any untransformable edge pixels
515 rather than having a strange-looking strip along the right and/or bottom edges
516 of a transformed image. To do this, add the -trim switch:
517 -trim Drop non-transformable edge blocks.
518 Obviously, a transformation with -trim is not reversible, so strictly speaking
519 jpegtran with this switch is not lossless. Also, the expected mathematical
520 equivalences between the transformations no longer hold. For example,
521 "-rot 270 -trim" trims only the bottom edge, but "-rot 90 -trim" followed by
522 "-rot 180 -trim" trims both edges.
524 If you are only interested in perfect transformations, add the -perfect switch:
525 -perfect Fail with an error if the transformation is not
527 For example, you may want to do
528 jpegtran -rot 90 -perfect foo.jpg || djpeg foo.jpg | pnmflip -r90 | cjpeg
529 to do a perfect rotation, if available, or an approximated one if not.
531 This version of jpegtran also offers a lossless crop option, which discards
532 data outside of a given image region but losslessly preserves what is inside.
533 Like the rotate and flip transforms, lossless crop is restricted by the current
534 JPEG format; the upper left corner of the selected region must fall on an iMCU
535 boundary. If it doesn't, then it is silently moved up and/or left to the
536 nearest iMCU boundary (the lower right corner is unchanged.) Thus, the output
537 image covers at least the requested region, but it may cover more. The
538 adjustment of the region dimensions may be optionally disabled by attaching an
539 'f' character ("force") to the width or height number.
541 The image can be losslessly cropped by giving the switch:
542 -crop WxH+X+Y Crop to a rectangular region of width W and height H,
543 starting at point X,Y.
545 If W or H is larger than the width/height of the input image, then the output
546 image is expanded in size, and the expanded region is filled in with zeros
547 (neutral gray). Attaching an 'f' character ("flatten") to the width number
548 will cause each block in the expanded region to be filled in with the DC
549 coefficient of the nearest block in the input image rather than grayed out.
550 Attaching an 'r' character ("reflect") to the width number will cause the
551 expanded region to be filled in with repeated reflections of the input image
552 rather than grayed out.
554 A complementary lossless wipe option is provided to discard (gray out) data
555 inside a given image region while losslessly preserving what is outside:
556 -wipe WxH+X+Y Wipe (gray out) a rectangular region of width W and
557 height H from the input image, starting at point X,Y.
559 Attaching an 'f' character ("flatten") to the width number will cause the
560 region to be filled with the average of adjacent blocks rather than grayed out.
561 If the wipe region and the region outside the wipe region, when adjusted to the
562 nearest iMCU boundary, form two horizontally adjacent rectangles, then
563 attaching an 'r' character ("reflect") to the width number will cause the wipe
564 region to be filled with repeated reflections of the outside region rather than
567 A lossless drop option is also provided, which allows another JPEG image to be
568 inserted ("dropped") into the input image data at a given position, replacing
569 the existing image data at that position:
570 -drop +X+Y filename Drop (insert) another image at point X,Y
572 Both the input image and the drop image must have the same subsampling level.
573 It is best if they also have the same quantization (quality.) Otherwise, the
574 quantization of the output image will be adapted to accommodate the higher of
575 the input image quality and the drop image quality. The trim option can be
576 used with the drop option to requantize the drop image to match the input
577 image. Note that a grayscale image can be dropped into a full-color image or
578 vice versa, as long as the full-color image has no vertical subsampling. If
579 the input image is grayscale and the drop image is full-color, then the
580 chrominance channels from the drop image will be discarded.
582 Other not-strictly-lossless transformation switches are:
584 -grayscale Force grayscale output.
585 This option discards the chrominance channels if the input image is YCbCr
586 (ie, a standard color JPEG), resulting in a grayscale JPEG file. The
587 luminance channel is preserved exactly, so this is a better method of reducing
588 to grayscale than decompression, conversion, and recompression. This switch
589 is particularly handy for fixing a monochrome picture that was mistakenly
590 encoded as a color JPEG. (In such a case, the space savings from getting rid
591 of the near-empty chroma channels won't be large; but the decoding time for
592 a grayscale JPEG is substantially less than that for a color JPEG.)
594 jpegtran also recognizes these switches that control what to do with "extra"
595 markers, such as comment blocks:
596 -copy none Copy no extra markers from source file. This setting
597 suppresses all comments and other metadata in the
599 -copy comments Copy only comment markers. This setting copies
600 comments from the source file but discards any other
602 -copy icc Copy only ICC profile markers. This setting copies the
603 ICC profile from the source file but discards any other
605 -copy all Copy all extra markers. This setting preserves
606 miscellaneous markers found in the source file, such
607 as JFIF thumbnails, Exif data, and Photoshop settings.
608 In some files, these extra markers can be sizable.
609 Note that this option will copy thumbnails as-is;
610 they will not be transformed.
611 The default behavior is -copy comments. (Note: in IJG releases v6 and v6a,
612 jpegtran always did the equivalent of -copy none.)
614 Additional switches recognized by jpegtran are:
619 These work the same as in cjpeg or djpeg.
622 THE COMMENT UTILITIES
624 The JPEG standard allows "comment" (COM) blocks to occur within a JPEG file.
625 Although the standard doesn't actually define what COM blocks are for, they
626 are widely used to hold user-supplied text strings. This lets you add
627 annotations, titles, index terms, etc to your JPEG files, and later retrieve
628 them as text. COM blocks do not interfere with the image stored in the JPEG
629 file. The maximum size of a COM block is 64K, but you can have as many of
630 them as you like in one JPEG file.
632 We provide two utility programs to display COM block contents and add COM
633 blocks to a JPEG file.
635 rdjpgcom searches a JPEG file and prints the contents of any COM blocks on
636 standard output. The command line syntax is
637 rdjpgcom [-raw] [-verbose] [inputfilename]
638 The switch "-raw" (or just "-r") causes rdjpgcom to output non-printable
639 characters in JPEG comments. These characters are normally escaped for
641 The switch "-verbose" (or just "-v") causes rdjpgcom to also display the JPEG
642 image dimensions. If you omit the input file name from the command line,
643 the JPEG file is read from standard input. (This may not work on some
644 operating systems, if binary data can't be read from stdin.)
646 wrjpgcom adds a COM block, containing text you provide, to a JPEG file.
647 Ordinarily, the COM block is added after any existing COM blocks, but you
648 can delete the old COM blocks if you wish. wrjpgcom produces a new JPEG
649 file; it does not modify the input file. DO NOT try to overwrite the input
650 file by directing wrjpgcom's output back into it; on most systems this will
651 just destroy your file.
653 The command line syntax for wrjpgcom is similar to cjpeg's. On most systems,
655 wrjpgcom [switches] [inputfilename]
656 The output file is written to standard output. The input file comes from
657 the named file, or from standard input if no input file is named.
659 If you defined TWO_FILE_COMMANDLINE when compiling the program, the syntax is:
660 wrjpgcom [switches] inputfilename outputfilename
661 where both input and output file names must be given explicitly.
663 wrjpgcom understands three switches:
664 -replace Delete any existing COM blocks from the file.
665 -comment "Comment text" Supply new COM text on command line.
666 -cfile name Read text for new COM block from named file.
667 (Switch names can be abbreviated.) If you have only one line of comment text
668 to add, you can provide it on the command line with -comment. The comment
669 text must be surrounded with quotes so that it is treated as a single
670 argument. Longer comments can be read from a text file.
672 If you give neither -comment nor -cfile, then wrjpgcom will read the comment
673 text from standard input. (In this case an input image file name MUST be
674 supplied, so that the source JPEG file comes from somewhere else.) You can
675 enter multiple lines, up to 64KB worth. Type an end-of-file indicator
676 (usually control-D or control-Z) to terminate the comment text entry.
678 wrjpgcom will not add a COM block if the provided comment string is empty.
679 Therefore -replace -comment "" can be used to delete all COM blocks from a
682 These utility programs do not depend on the IJG JPEG library. In
683 particular, the source code for rdjpgcom is intended as an illustration of
684 the minimum amount of code required to parse a JPEG file header correctly.