1CJPEG(1) General Commands Manual CJPEG(1)
2
6 cjpeg - compress an image file to a JPEG file
7
9 cjpeg [ options ] [ filename ]
10
12 cjpeg compresses the named image file, or the standard input if no file
13 is named, and produces a JPEG/JFIF file on the standard output. The
14 currently supported input file formats are: PPM (PBMPLUS color format),
15 PGM (PBMPLUS grayscale format), BMP, GIF, and Targa.
16
18 All switch names may be abbreviated; for example, -grayscale may be
19 written -gray or -gr. Most of the "basic" switches can be abbreviated
20 to as little as one letter. Upper and lower case are equivalent (thus
21 -BMP is the same as -bmp). British spellings are also accepted (e.g.,
22 -greyscale), though for brevity these are not mentioned below.
23 The basic switches are:
25 -quality N[,...]
27 Scale quantization tables to adjust image quality. Quality is 0
28 (worst) to 100 (best); default is 75. (See below for more
29 info.)
30 -grayscale
32 Create monochrome JPEG file from color input. Be sure to use
33 this switch when compressing a grayscale BMP or GIF file, be‐
34 cause cjpeg isn't bright enough to notice whether a BMP or GIF
35 file uses only shades of gray. By saying -grayscale, you'll get
36 a smaller JPEG file that takes less time to process.
37 -rgb Create RGB JPEG file. Using this switch suppresses the conver‐
39 sion from RGB colorspace input to the default YCbCr JPEG col‐
40 orspace.
41 -optimize
43 Perform optimization of entropy encoding parameters. Without
44 this, default encoding parameters are used. -optimize usually
45 makes the JPEG file a little smaller, but cjpeg runs somewhat
46 slower and needs much more memory. Image quality and speed of
47 decompression are unaffected by -optimize.
48 -progressive
50 Create progressive JPEG file (see below).
51 -targa Input file is Targa format. Targa files that contain an "iden‐
53 tification" field will not be automatically recognized by cjpeg;
54 for such files you must specify -targa to make cjpeg treat the
55 input as Targa format. For most Targa files, you won't need
56 this switch.
57 The -quality switch lets you trade off compressed file size against
59 quality of the reconstructed image: the higher the quality setting, the
60 larger the JPEG file, and the closer the output image will be to the
61 original input. Normally you want to use the lowest quality setting
62 (smallest file) that decompresses into something visually indistin‐
63 guishable from the original image. For this purpose the quality set‐
64 ting should generally be between 50 and 95 (the default is 75) for pho‐
65 tographic images. If you see defects at -quality 75, then go up 5 or
66 10 counts at a time until you are happy with the output image. (The
67 optimal setting will vary from one image to another.)
68 -quality 100 will generate a quantization table of all 1's, minimizing
70 loss in the quantization step (but there is still information loss in
71 subsampling, as well as roundoff error.) For most images, specifying a
72 quality value above about 95 will increase the size of the compressed
73 file dramatically, and while the quality gain from these higher quality
74 values is measurable (using metrics such as PSNR or SSIM), it is rarely
75 perceivable by human vision.
76 In the other direction, quality values below 50 will produce very small
78 files of low image quality. Settings around 5 to 10 might be useful in
79 preparing an index of a large image library, for example. Try -quality
80 2 (or so) for some amusing Cubist effects. (Note: quality values below
81 about 25 generate 2-byte quantization tables, which are considered op‐
82 tional in the JPEG standard. cjpeg emits a warning message when you
83 give such a quality value, because some other JPEG programs may be un‐
84 able to decode the resulting file. Use -baseline if you need to ensure
85 compatibility at low quality values.)
86 The -quality option has been extended in this version of cjpeg to sup‐
88 port separate quality settings for luminance and chrominance (or, in
89 general, separate settings for every quantization table slot.) The
90 principle is the same as chrominance subsampling: since the human eye
91 is more sensitive to spatial changes in brightness than spatial changes
92 in color, the chrominance components can be quantized more than the lu‐
93 minance components without incurring any visible image quality loss.
94 However, unlike subsampling, this feature reduces data in the frequency
95 domain instead of the spatial domain, which allows for more fine-
96 grained control. This option is useful in quality-sensitive applica‐
97 tions, for which the artifacts generated by subsampling may be unac‐
98 ceptable.
99 The -quality option accepts a comma-separated list of parameters, which
101 respectively refer to the quality levels that should be assigned to the
102 quantization table slots. If there are more q-table slots than parame‐
103 ters, then the last parameter is replicated. Thus, if only one quality
104 parameter is given, this is used for both luminance and chrominance
105 (slots 0 and 1, respectively), preserving the legacy behavior of cjpeg
106 v6b and prior. More (or customized) quantization tables can be set
107 with the -qtables option and assigned to components with the -qslots
108 option (see the "wizard" switches below.)
109 JPEG files generated with separate luminance and chrominance quality
111 are fully compliant with standard JPEG decoders.
112 CAUTION: For this setting to be useful, be sure to pass an argument of
114 -sample 1x1 to cjpeg to disable chrominance subsampling. Otherwise,
115 the default subsampling level (2x2, AKA "4:2:0") will be used.
116 The -progressive switch creates a "progressive JPEG" file. In this
118 type of JPEG file, the data is stored in multiple scans of increasing
119 quality. If the file is being transmitted over a slow communications
120 link, the decoder can use the first scan to display a low-quality image
121 very quickly, and can then improve the display with each subsequent
122 scan. The final image is exactly equivalent to a standard JPEG file of
123 the same quality setting, and the total file size is about the same ---
124 often a little smaller.
125 Switches for advanced users:
127 -arithmetic
129 Use arithmetic coding. Caution: arithmetic coded JPEG is not
130 yet widely implemented, so many decoders will be unable to view
131 an arithmetic coded JPEG file at all.
132 -dct int
134 Use accurate integer DCT method (default).
135 -dct fast
137 Use less accurate integer DCT method [legacy feature]. When the
138 Independent JPEG Group's software was first released in 1991,
139 the compression time for a 1-megapixel JPEG image on a main‐
140 stream PC was measured in minutes. Thus, the fast integer DCT
141 algorithm provided noticeable performance benefits. On modern
142 CPUs running libjpeg-turbo, however, the compression time for a
143 1-megapixel JPEG image is measured in milliseconds, and thus the
144 performance benefits of the fast algorithm are much less notice‐
145 able. On modern x86/x86-64 CPUs that support AVX2 instructions,
146 the fast and int methods have similar performance. On other
147 types of CPUs, the fast method is generally about 5-15% faster
148 than the int method.
149 For quality levels of 90 and below, there should be little or no
151 perceptible quality difference between the two algorithms. For
152 quality levels above 90, however, the difference between the
153 fast and int methods becomes more pronounced. With quality=97,
154 for instance, the fast method incurs generally about a 1-3 dB
155 loss in PSNR relative to the int method, but this can be larger
156 for some images. Do not use the fast method with quality levels
157 above 97. The algorithm often degenerates at quality=98 and
158 above and can actually produce a more lossy image than if lower
159 quality levels had been used. Also, in libjpeg-turbo, the fast
160 method is not fully accelerated for quality levels above 97, so
161 it will be slower than the int method.
162 -dct float
164 Use floating-point DCT method [legacy feature]. The float
165 method does not produce significantly more accurate results than
166 the int method, and it is much slower. The float method may
167 also give different results on different machines due to varying
168 roundoff behavior, whereas the integer methods should give the
169 same results on all machines.
170 -icc file
172 Embed ICC color management profile contained in the specified
173 file.
174 -restart N
176 Emit a JPEG restart marker every N MCU rows, or every N MCU
177 blocks if "B" is attached to the number. -restart 0 (the de‐
178 fault) means no restart markers.
179 -smooth N
181 Smooth the input image to eliminate dithering noise. N, ranging
182 from 1 to 100, indicates the strength of smoothing. 0 (the de‐
183 fault) means no smoothing.
184 -maxmemory N
186 Set limit for amount of memory to use in processing large im‐
187 ages. Value is in thousands of bytes, or millions of bytes if
188 "M" is attached to the number. For example, -max 4m selects
189 4000000 bytes. If more space is needed, an error will occur.
190 -outfile name
192 Send output image to the named file, not to standard output.
193 -memdst
195 Compress to memory instead of a file. This feature was imple‐
196 mented mainly as a way of testing the in-memory destination man‐
197 ager (jpeg_mem_dest()), but it is also useful for benchmarking,
198 since it reduces the I/O overhead.
199 -report
201 Report compression progress.
202 -strict
204 Treat all warnings as fatal. Enabling this option will cause
205 the compressor to abort if an LZW-compressed GIF input image
206 contains incomplete or corrupt image data.
207 -verbose
209 Enable debug printout. More -v's give more output. Also, ver‐
210 sion information is printed at startup.
211 -debug Same as -verbose.
213 -version
215 Print version information and exit.
216 The -restart option inserts extra markers that allow a JPEG decoder to
218 resynchronize after a transmission error. Without restart markers, any
219 damage to a compressed file will usually ruin the image from the point
220 of the error to the end of the image; with restart markers, the damage
221 is usually confined to the portion of the image up to the next restart
222 marker. Of course, the restart markers occupy extra space. We recom‐
223 mend -restart 1 for images that will be transmitted across unreliable
224 networks such as Usenet.
225 The -smooth option filters the input to eliminate fine-scale noise.
227 This is often useful when converting dithered images to JPEG: a moder‐
228 ate smoothing factor of 10 to 50 gets rid of dithering patterns in the
229 input file, resulting in a smaller JPEG file and a better-looking im‐
230 age. Too large a smoothing factor will visibly blur the image, how‐
231 ever.
232 Switches for wizards:
234 -baseline
236 Force baseline-compatible quantization tables to be generated.
237 This clamps quantization values to 8 bits even at low quality
238 settings. (This switch is poorly named, since it does not en‐
239 sure that the output is actually baseline JPEG. For example,
240 you can use -baseline and -progressive together.)
241 -qtables file
243 Use the quantization tables given in the specified text file.
244 -qslots N[,...]
246 Select which quantization table to use for each color component.
247 -sample HxV[,...]
249 Set JPEG sampling factors for each color component.
250 -scans file
252 Use the scan script given in the specified text file.
253 The "wizard" switches are intended for experimentation with JPEG. If
255 you don't know what you are doing, don't use them. These switches are
256 documented further in the file wizard.txt.
257
259 This example compresses the PPM file foo.ppm with a quality factor of
260 60 and saves the output as foo.jpg:
261 cjpeg -quality 60 foo.ppm > foo.jpg
263
265 Color GIF files are not the ideal input for JPEG; JPEG is really in‐
266 tended for compressing full-color (24-bit) images. In particular,
267 don't try to convert cartoons, line drawings, and other images that
268 have only a few distinct colors. GIF works great on these, JPEG does
269 not. If you want to convert a GIF to JPEG, you should experiment with
270 cjpeg's -quality and -smooth options to get a satisfactory conversion.
271 -smooth 10 or so is often helpful.
272 Avoid running an image through a series of JPEG compression/decompres‐
274 sion cycles. Image quality loss will accumulate; after ten or so cy‐
275 cles the image may be noticeably worse than it was after one cycle.
276 It's best to use a lossless format while manipulating an image, then
277 convert to JPEG format when you are ready to file the image away.
278 The -optimize option to cjpeg is worth using when you are making a "fi‐
280 nal" version for posting or archiving. It's also a win when you are
281 using low quality settings to make very small JPEG files; the percent‐
282 age improvement is often a lot more than it is on larger files. (At
283 present, -optimize mode is always selected when generating progressive
284 JPEG files.)
285
287 JPEGMEM
288 If this environment variable is set, its value is the default
289 memory limit. The value is specified as described for the
290 -maxmemory switch. JPEGMEM overrides the default value speci‐
291 fied when the program was compiled, and itself is overridden by
292 an explicit -maxmemory.
293
295 djpeg(1), jpegtran(1), rdjpgcom(1), wrjpgcom(1)
296 ppm(5), pgm(5)
297 Wallace, Gregory K. "The JPEG Still Picture Compression Standard",
298 Communications of the ACM, April 1991 (vol. 34, no. 4), pp. 30-44.
299
301 Independent JPEG Group
302 This file was modified by The libjpeg-turbo Project to include only in‐
304 formation relevant to libjpeg-turbo, to wordsmith certain sections, and
305 to describe features not present in libjpeg.
306
308 Not all variants of BMP and Targa file formats are supported.
309 The -targa switch is not a bug, it's a feature. (It would be a bug if
311 the Targa format designers had not been clueless.)
312 18 November 2021 CJPEG(1)