1CWEBP(1) General Commands Manual CWEBP(1)
2
6 cwebp - compress an image file to a WebP file
7
9 cwebp [options] input_file -o output_file.webp
10
12 This manual page documents the cwebp command.
13 cwebp compresses an image using the WebP format. Input format can be
15 either PNG, JPEG, TIFF, WebP or raw Y'CbCr samples.
16
18 The basic options are:
19 -o string
21 Specify the name of the output WebP file. If omitted, cwebp will
22 perform compression but only report statistics. Using "-" as
23 output name will direct output to 'stdout'.
24 -- string
26 Explicitly specify the input file. This option is useful if the
27 input file starts with a '-' for instance. This option must
28 appear last. Any other options afterward will be ignored.
29 -h, -help
31 A short usage summary.
32 -H, -longhelp
34 A summary of all the possible options.
35 -version
37 Print the version number (as major.minor.revision) and exit.
38 -lossless
40 Encode the image without any loss. For images with fully trans‐
41 parent area, the invisible pixel values (R/G/B or Y/U/V) will be
42 preserved only if the -exact option is used.
43 -near_lossless int
45 Specify the level of near-lossless image preprocessing. This
46 option adjusts pixel values to help compressibility, but has
47 minimal impact on the visual quality. It triggers lossless com‐
48 pression mode automatically. The range is 0 (maximum preprocess‐
49 ing) to 100 (no preprocessing, the default). The typical value
50 is around 60. Note that lossy with -q 100 can at times yield
51 better results.
52 -q float
54 Specify the compression factor for RGB channels between 0 and
55 100. The default is 75.
56 In case of lossy compression (default), a small factor produces
57 a smaller file with lower quality. Best quality is achieved by
58 using a value of 100.
59 In case of lossless compression (specified by the -lossless
60 option), a small factor enables faster compression speed, but
61 produces a larger file. Maximum compression is achieved by
62 using a value of 100.
63 -z int Switch on lossless compression mode with the specified level
65 between 0 and 9, with level 0 being the fastest, 9 being the
66 slowest. Fast mode produces larger file size than slower ones. A
67 good default is -z 6. This option is actually a shortcut for
68 some predefined settings for quality and method. If options -q
69 or -m are subsequently used, they will invalidate the effect of
70 this option.
71 -alpha_q int
73 Specify the compression factor for alpha compression between 0
74 and 100. Lossless compression of alpha is achieved using a
75 value of 100, while the lower values result in a lossy compres‐
76 sion. The default is 100.
77 -preset string
79 Specify a set of pre-defined parameters to suit a particular
80 type of source material. Possible values are: default, photo,
81 picture, drawing, icon, text. Since -preset overwrites the other
82 parameters' values (except the -q one), this option should
83 preferably appear first in the order of the arguments.
84 -m int Specify the compression method to use. This parameter controls
86 the trade off between encoding speed and the compressed file
87 size and quality. Possible values range from 0 to 6. Default
88 value is 4. When higher values are used, the encoder will spend
89 more time inspecting additional encoding possibilities and
90 decide on the quality gain. Lower value can result in faster
91 processing time at the expense of larger file size and lower
92 compression quality.
93 -resize width height
95 Resize the source to a rectangle with size width x height. If
96 either (but not both) of the width or height parameters is 0,
97 the value will be calculated preserving the aspect-ratio.
98 -crop x_position y_position width height
100 Crop the source to a rectangle with top-left corner at coordi‐
101 nates (x_position, y_position) and size width x height. This
102 cropping area must be fully contained within the source rectan‐
103 gle.
104 -mt Use multi-threading for encoding, if possible.
106 -low_memory
108 Reduce memory usage of lossy encoding by saving four times the
109 compressed size (typically). This will make the encoding slower
110 and the output slightly different in size and distortion. This
111 flag is only effective for methods 3 and up, and is off by
112 default. Note that leaving this flag off will have some side
113 effects on the bitstream: it forces certain bitstream features
114 like number of partitions (forced to 1). Note that a more
115 detailed report of bitstream size is printed by cwebp when using
116 this option.
117 LOSSY OPTIONS
120 These options are only effective when doing lossy encoding (the
121 default, with or without alpha).
122 -size int
125 Specify a target size (in bytes) to try and reach for the com‐
126 pressed output. The compressor will make several passes of par‐
127 tial encoding in order to get as close as possible to this tar‐
128 get. If both -size and -psnr are used, -size value will prevail.
129 -psnr float
131 Specify a target PSNR (in dB) to try and reach for the com‐
132 pressed output. The compressor will make several passes of par‐
133 tial encoding in order to get as close as possible to this tar‐
134 get. If both -size and -psnr are used, -size value will prevail.
135 -pass int
137 Set a maximum number of passes to use during the dichotomy used
138 by options -size or -psnr. Maximum value is 10, default is 1.
139 If options -size or -psnr were used, but -pass wasn't specified,
140 a default value of '6' passes will be used.
141 -af Turns auto-filter on. This algorithm will spend additional time
143 optimizing the filtering strength to reach a well-balanced qual‐
144 ity.
145 -jpeg_like
147 Change the internal parameter mapping to better match the
148 expected size of JPEG compression. This flag will generally pro‐
149 duce an output file of similar size to its JPEG equivalent (for
150 the same -q setting), but with less visual distortion.
151 Advanced options:
154 -f int Specify the strength of the deblocking filter, between 0 (no
157 filtering) and 100 (maximum filtering). A value of 0 will turn
158 off any filtering. Higher value will increase the strength of
159 the filtering process applied after decoding the picture. The
160 higher the value the smoother the picture will appear. Typical
161 values are usually in the range of 20 to 50.
162 -sharpness int
164 Specify the sharpness of the filtering (if used). Range is 0
165 (sharpest) to 7 (least sharp). Default is 0.
166 -strong
168 Use strong filtering (if filtering is being used thanks to the
169 -f option). Strong filtering is on by default.
170 -nostrong
172 Disable strong filtering (if filtering is being used thanks to
173 the -f option) and use simple filtering instead.
174 -sharp_yuv
176 Use more accurate and sharper RGB->YUV conversion if needed.
177 Note that this process is slower than the default 'fast'
178 RGB->YUV conversion.
179 -sns int
181 Specify the amplitude of the spatial noise shaping. Spatial
182 noise shaping (or sns for short) refers to a general collection
183 of built-in algorithms used to decide which area of the picture
184 should use relatively less bits, and where else to better trans‐
185 fer these bits. The possible range goes from 0 (algorithm is
186 off) to 100 (the maximal effect). The default value is 50.
187 -segments int
189 Change the number of partitions to use during the segmentation
190 of the sns algorithm. Segments should be in range 1 to 4.
191 Default value is 4. This option has no effect for methods 3 and
192 up, unless -low_memory is used.
193 -partition_limit int
195 Degrade quality by limiting the number of bits used by some mac‐
196 roblocks. Range is 0 (no degradation, the default) to 100 (full
197 degradation). Useful values are usually around 30-70 for moder‐
198 ately large images. In the VP8 format, the so-called control
199 partition has a limit of 512k and is used to store the following
200 information: whether the macroblock is skipped, which segment it
201 belongs to, whether it is coded as intra 4x4 or intra 16x16
202 mode, and finally the prediction modes to use for each of the
203 sub-blocks. For a very large image, 512k only leaves room to
204 few bits per 16x16 macroblock. The absolute minimum is 4 bits
205 per macroblock. Skip, segment, and mode information can use up
206 almost all these 4 bits (although the case is unlikely), which
207 is problematic for very large images. The partition_limit factor
208 controls how frequently the most bit-costly mode (intra 4x4)
209 will be used. This is useful in case the 512k limit is reached
210 and the following message is displayed: Error code: 6 (PARTI‐
211 TION0_OVERFLOW: Partition #0 is too big to fit 512k). If using
212 -partition_limit is not enough to meet the 512k constraint, one
213 should use less segments in order to save more header bits per
214 macroblock. See the -segments option.
215 LOGGING OPTIONS
218 These options control the level of output:
219 -v Print extra information (encoding time in particular).
221 -print_psnr
223 Compute and report average PSNR (Peak-Signal-To-Noise ratio).
224 -print_ssim
226 Compute and report average SSIM (structural similarity metric,
227 see http://en.wikipedia.org/wiki/SSIM for additional details).
228 -print_lsim
230 Compute and report local similarity metric (sum of lowest error
231 amongst the collocated pixel neighbors).
232 -progress
234 Report encoding progress in percent.
235 -quiet Do not print anything.
237 -short Only print brief information (output file size and PSNR) for
239 testing purposes.
240 -map int
242 Output additional ASCII-map of encoding information. Possible
243 map values range from 1 to 6. This is only meant to help debug‐
244 ging.
245 ADDITIONAL OPTIONS
248 More advanced options are:
249 -s width height
251 Specify that the input file actually consists of raw Y'CbCr sam‐
252 ples following the ITU-R BT.601 recommendation, in 4:2:0 linear
253 format. The luma plane has size width x height.
254 -pre int
256 Specify some preprocessing steps. Using a value of '2' will
257 trigger quality-dependent pseudo-random dithering during
258 RGBA->YUVA conversion (lossy compression only).
259 -alpha_filter string
261 Specify the predictive filtering method for the alpha plane. One
262 of 'none', 'fast' or 'best', in increasing complexity and slow‐
263 ness order. Default is 'fast'. Internally, alpha filtering is
264 performed using four possible predictions (none, horizontal,
265 vertical, gradient). The 'best' mode will try each mode in turn
266 and pick the one which gives the smaller size. The 'fast' mode
267 will just try to form an a priori guess without testing all
268 modes.
269 -alpha_method int
271 Specify the algorithm used for alpha compression: 0 or 1. Algo‐
272 rithm 0 denotes no compression, 1 uses WebP lossless format for
273 compression. The default is 1.
274 -exact Preserve RGB values in transparent area. The default is off, to
276 help compressibility.
277 -blend_alpha int
279 This option blends the alpha channel (if present) with the
280 source using the background color specified in hexadecimal as
281 0xrrggbb. The alpha channel is afterward reset to the opaque
282 value 255.
283 -noalpha
285 Using this option will discard the alpha channel.
286 -hint string
288 Specify the hint about input image type. Possible values are:
289 photo, picture or graph.
290 -metadata string
292 A comma separated list of metadata to copy from the input to the
293 output if present. Valid values: all, none, exif, icc, xmp.
294 The default is none.
295 Note: each input format may not support all combinations.
297 -noasm Disable all assembly optimizations.
299
302 Please report all bugs to the issue tracker:
303 https://bugs.chromium.org/p/webp
304 Patches welcome! See this page to get started: http://www.webmpro‐
305 ject.org/code/contribute/submitting-patches/
306
309 cwebp -q 50 -lossless picture.png -o picture_lossless.webp
310 cwebp -q 70 picture_with_alpha.png -o picture_with_alpha.webp
311 cwebp -sns 70 -f 50 -size 60000 picture.png -o picture.webp
312 cwebp -o picture.webp -- ---picture.png
313
316 cwebp is a part of libwebp and was written by the WebP team.
317 The latest source tree is available at https://chromium.google‐
318 source.com/webm/libwebp
319 This manual page was written by Pascal Massimino <pascal.mas‐
321 simino@gmail.com>, for the Debian project (and may be used by others).
322
325 dwebp(1), gif2webp(1)
326 Please refer to http://developers.google.com/speed/webp/ for additional
327 information.
328 January 11, 2019 CWEBP(1)