1 .TH JPEGTRAN 1 "28 December 2009"
3 jpegtran \- lossless transformation of JPEG files
16 performs various useful transformations of JPEG files.
17 It can translate the coded representation from one variant of JPEG to another,
18 for example from baseline JPEG to progressive JPEG or vice versa. It can also
19 perform some rearrangements of the image data, for example turning an image
20 from landscape to portrait format by rotation.
23 works by rearranging the compressed data (DCT coefficients), without
24 ever fully decoding the image. Therefore, its transformations are lossless:
25 there is no image degradation at all, which would not be true if you used
29 to accomplish the same conversion. But by the same token,
31 cannot perform lossy operations such as changing the image quality.
34 reads the named JPEG/JFIF file, or the standard input if no file is
35 named, and produces a JPEG/JFIF file on the standard output.
37 All switch names may be abbreviated; for example,
43 Upper and lower case are equivalent.
44 British spellings are also accepted (e.g.,
46 though for brevity these are not mentioned below.
48 To specify the coded JPEG representation used in the output file,
50 accepts a subset of the switches recognized by
54 Perform optimization of entropy encoding parameters.
57 Create progressive JPEG file.
60 Emit a JPEG restart marker every N MCU rows, or every N MCU blocks if "B" is
61 attached to the number.
64 Use arithmetic coding.
67 Use the scan script given in the specified text file.
71 for more details about these switches.
72 If you specify none of these switches, you get a plain baseline-JPEG output
73 file. The quality setting and so forth are determined by the input file.
75 The image can be losslessly transformed by giving one of these switches:
78 Mirror image horizontally (left-right).
81 Mirror image vertically (top-bottom).
84 Rotate image 90 degrees clockwise.
87 Rotate image 180 degrees.
90 Rotate image 270 degrees clockwise (or 90 ccw).
93 Transpose image (across UL-to-LR axis).
96 Transverse transpose (across UR-to-LL axis).
98 The transpose transformation has no restrictions regarding image dimensions.
99 The other transformations operate rather oddly if the image dimensions are not
100 a multiple of the iMCU size (usually 8 or 16 pixels), because they can only
101 transform complete blocks of DCT coefficient data in the desired way.
104 default behavior when transforming an odd-size image is designed
105 to preserve exact reversibility and mathematical consistency of the
106 transformation set. As stated, transpose is able to flip the entire image
107 area. Horizontal mirroring leaves any partial iMCU column at the right edge
108 untouched, but is able to flip all rows of the image. Similarly, vertical
109 mirroring leaves any partial iMCU row at the bottom edge untouched, but is
110 able to flip all columns. The other transforms can be built up as sequences
111 of transpose and flip operations; for consistency, their actions on edge
112 pixels are defined to be the same as the end result of the corresponding
113 transpose-and-flip sequence.
115 For practical use, you may prefer to discard any untransformable edge pixels
116 rather than having a strange-looking strip along the right and/or bottom edges
117 of a transformed image. To do this, add the
122 Drop non-transformable edge blocks.
124 Obviously, a transformation with
126 is not reversible, so strictly speaking
128 with this switch is not lossless. Also, the expected mathematical
129 equivalences between the transformations no longer hold. For example,
131 trims only the bottom edge, but
137 If you are only interested in perfect transformation, add the
142 Fails with an error if the transformation is not perfect.
144 For example you may want to do
146 .B (jpegtran \-rot 90 -perfect
150 .B | pnmflip \-r90 | cjpeg)
152 to do a perfect rotation if available or an approximated one if not.
154 We also offer a lossless-crop option, which discards data outside a given
155 image region but losslessly preserves what is inside. Like the rotate and
156 flip transforms, lossless crop is restricted by the current JPEG format: the
157 upper left corner of the selected region must fall on an iMCU boundary. If
158 this does not hold for the given crop parameters, we silently move the upper
159 left corner up and/or left to make it so, simultaneously increasing the region
160 dimensions to keep the lower right crop corner unchanged. (Thus, the output
161 image covers at least the requested region, but may cover more.)
163 The image can be losslessly cropped by giving the switch:
166 Crop to a rectangular subarea of width W, height H starting at point X,Y.
168 Other not-strictly-lossless transformation switches are:
171 Force grayscale output.
173 This option discards the chrominance channels if the input image is YCbCr
174 (ie, a standard color JPEG), resulting in a grayscale JPEG file. The
175 luminance channel is preserved exactly, so this is a better method of reducing
176 to grayscale than decompression, conversion, and recompression. This switch
177 is particularly handy for fixing a monochrome picture that was mistakenly
178 encoded as a color JPEG. (In such a case, the space savings from getting rid
179 of the near-empty chroma channels won't be large; but the decoding time for
180 a grayscale JPEG is substantially less than that for a color JPEG.)
183 Scale the output image by a factor M/N.
185 Currently supported scale factors are M/N with all M from 1 to 16, where N is
186 the source DCT size, which is 8 for baseline JPEG. If the /N part is omitted,
187 then M specifies the DCT scaled size to be applied on the given input. For
188 baseline JPEG this is equivalent to M/8 scaling, since the source DCT size
189 for baseline JPEG is 8.
191 An implementation of the JPEG SmartScale extension is required for this
192 feature. SmartScale enabled JPEG is not yet widely implemented, so many
193 decoders will be unable to view a SmartScale extended JPEG file at all.
196 also recognizes these switches that control what to do with "extra" markers,
197 such as comment blocks:
200 Copy no extra markers from source file. This setting suppresses all
201 comments and other excess baggage present in the source file.
204 Copy only comment markers. This setting copies comments from the source file,
205 but discards any other inessential (for image display) data.
208 Copy all extra markers. This setting preserves miscellaneous markers
209 found in the source file, such as JFIF thumbnails, Exif data, and Photoshop
210 settings. In some files these extra markers can be sizable.
212 The default behavior is
213 .BR "\-copy comments" .
214 (Note: in IJG releases v6 and v6a,
216 always did the equivalent of
219 Additional switches recognized by jpegtran are:
222 Set limit for amount of memory to use in processing large images. Value is
223 in thousands of bytes, or millions of bytes if "M" is attached to the
226 selects 4000000 bytes. If more space is needed, temporary files will be used.
228 .BI \-outfile " name"
229 Send output image to the named file, not to standard output.
232 Enable debug printout. More
234 give more output. Also, version information is printed at startup.
241 This example converts a baseline JPEG file to progressive form:
243 .B jpegtran \-progressive
248 This example rotates an image 90 degrees clockwise, discarding any
249 unrotatable edge pixels:
251 .B jpegtran \-rot 90 -trim
258 If this environment variable is set, its value is the default memory limit.
259 The value is specified as described for the
263 overrides the default value specified when the program was compiled, and
264 itself is overridden by an explicit
272 Wallace, Gregory K. "The JPEG Still Picture Compression Standard",
273 Communications of the ACM, April 1991 (vol. 34, no. 4), pp. 30-44.
275 Independent JPEG Group
277 The transform options can't transform odd-size images perfectly. Use
281 if you don't like the results.
283 The entire image is read into memory and then written out again, even in
284 cases where this isn't really necessary. Expect swapping on large images,
285 especially when using the more complex transform options.