This is a big headache for packaging jpeg-archive unless it has changes that cannot be merged upstream

You should add libtool as a dependency for installing on Ubuntu & Debian

$ sudo apt-get install build-essential autoconf nasm libtool

Hi,

I'm trying to compile jpeg-archive as part of a greater effort to get image_optim (https://github.com/toy/image_optim) working on FreeBSD (https://gist.github.com/murkfletcher/36f662a1c0ab59ea0eac), but I'm getting:

# make
make: "/usr/home/freebsd/jpeg-archive/Makefile" line 10: Missing dependency operator
make: "/usr/home/freebsd/jpeg-archive/Makefile" line 19: Need an operator
make: "/usr/home/freebsd/jpeg-archive/Makefile" line 31: Need an operator
make: Fatal errors encountered -- cannot continue
make: stopped in /usr/home/freebsd/jpeg-archive

Anybody know?

Thanks!

Is it possible to compress a file that is not written to the disk? Like, use it in an AWS lambda process so I could compress uploaded files? When you download a file from S3 to use it in lambda all i have is their buffer. I could write it to disk, then compress and upload but that would add processing time that could be saved.

thanks in advance.

I have an idea on optimization of JPEG Lossy

What is JPEG progressive? Actually it is a number of images where quality of each following image is higher previous the image.

We will take this image

JPEG Scan Killer project.

We receive 16 images:

How it is possible to create with the help -scans technology of optimization of JPEG lossy?

JPEG consists of blocks of 8х8 pixels. Perhaps, there is a sense to segment the image and to optimize each part separately. Later to connect parts of the image in the uniform optimized image. Principle of work of the betterjpeg program.

I wanted to learn, your opinion on the matter.

I want to compress an image into baseline JPEG.
What do you think about adding -b --baseline option to create baseline JPEG file (and disable progressive coding)?

Actually, mozjpeg supports -baseline option.

Hi!

When I have ran jpeg-recompress with the following paramters:

jpeg-recompress --accurate --ppm --method smallfry --strip --min 60 --max 100 input.ppm output.jpg

I got:

Final optimized smallfry at q=94: 101.582809

However, if I check the output with identify -verbose, it says that the quality of the output is only 85.
Also, when using other compression tools, I got a comparable file size at quality 85.
Has anybody experienced this issue?
With thanks,
Laszlo

Please allow more arguments to the "--subsample" switch.

Currently only 2x2 (default) and 1x1 (disable) is supported,
but there are plenty cameras available, that outputs 2x1 or 1x2 jpegs.
Transcoding these jpegs via jpeg-recompress are always too big (1x1),
or look often too ugly (2x2) (structure in red color areas, anti-aliasing, flat gradient etc.).

Sure, 2x2 is for the most jpegs enough quality, or simple not noticeable, so 2x2 can be still the default value.

So please, let us choose the correct sampling method, or autodetect it from the original.

Regards,
Biozynotiker

Would be quite nice to be able to disable output to stderr when optimizing images (if there isn't an error). Or write the information shown when optimizing to stdout and not to stderr.

Hi,
I have a few JPG that break jpeg recompress with I use the method smallfry. I am running the windows executable V2.1.1. No error message: just a windows popup that jpeg-recompress has stopped running.
It breaks just after displaying "Metadata size is xx kb".

If I use ssim as a method on the same files it's OK.
This happened on about 15 to 20% of tested files.

I've run into #16, and I was looking to upgrade my version of jpeg-archive to release v2.1.1 to see if it would fix the issue. I might be confused, but it seems that the Windows zip for v2.1.1 is missing jpeg-archive.sh, which was included with v2.1.0.

I believe this is not an issue in my case, since I just use jpeg-recompress directly, but I wanted to bring it to your attention in case the omission was accidental.

Thanks again for the great tool!

When I try to compress the following image with the smallfry method, jpeg-recompress crashes. I use the Windows binary (version 2.1.1).

http://i.imgur.com/YQYf8zc.jpg

Could the image dimensions 3000x1705 be the problem?

Hi!

1. I would like to have something like jpeg-recompress that maintain the perceived visual quality, but for WebP and also for PNG.

2. I'm also thinking of something not only like webp-recompress and png-recompress, but in addition something bigger,
   that could for example given a jpeg-recompress --auto-format option and a fixed SSIM,
   choose automatically between a JPEG or PNG8 based on the resulting file size and quality.

What do you think?

Hi,
followed your tutorial on how to compile on windows.
At the last command:

CC=gcc mingw32-make1

I get about 20 errors such as:
src/util.o:util.c:(.text+0x192): undefined reference to `jpeg_std_error'

and finally:
collect2.exe: error: ld returned 1 exit status

Please allow for export in PPM to allow using the matrix methods on different formats (JP2 or others)..

Thanks for nice tool.

It would be great if jpeg-compare could also read any lossless format like PNG or TIFF, using it as a "baseline". For example, I would like to compare the original image with two different derived JPGs. Now I have to convert PNG into JPG with quality=100 and use it as a "baseline".

If image-recompress fails to find a better quality for a given image, the resulting image is not encoded in progressive mode even if the option has been set.

The expected behavior is to get a progressive image as a result with the same quality.

I think the issue is that the Gentoo mozjpeg ebuild doesn't install include files... (while libjpeg-turbo does but app-archive doesn't use them).

$ make
cc -std=c99 -Wall -O3 -I/opt/mozjpeg/include -c -o src/util.o src/util.c
src/util.c: In function 'encodeJpeg':
src/util.c:145:13: warning: implicit declaration of function 'jpeg_c_int_param_supported' [-Wimplicit-function-declaration]
         if (jpeg_c_int_param_supported(&cinfo, JINT_COMPRESS_PROFILE)) {
             ^
src/util.c:145:48: error: 'JINT_COMPRESS_PROFILE' undeclared (first use in this function)
         if (jpeg_c_int_param_supported(&cinfo, JINT_COMPRESS_PROFILE)) {
                                                ^
src/util.c:145:48: note: each undeclared identifier is reported only once for each function it appears in
src/util.c:146:13: warning: implicit declaration of function 'jpeg_c_set_int_param' [-Wimplicit-function-declaration]
             jpeg_c_set_int_param(&cinfo, JINT_COMPRESS_PROFILE, JCP_FASTEST);
             ^
src/util.c:146:65: error: 'JCP_FASTEST' undeclared (first use in this function)
             jpeg_c_set_int_param(&cinfo, JINT_COMPRESS_PROFILE, JCP_FASTEST);
                                                                 ^
src/util.c:155:13: warning: implicit declaration of function 'jpeg_c_bool_param_supported' [-Wimplicit-function-declaration]
         if (jpeg_c_bool_param_supported(&cinfo, JBOOLEAN_TRELLIS_QUANT)) {
             ^
src/util.c:155:49: error: 'JBOOLEAN_TRELLIS_QUANT' undeclared (first use in this function)
         if (jpeg_c_bool_param_supported(&cinfo, JBOOLEAN_TRELLIS_QUANT)) {
                                                 ^
src/util.c:156:13: warning: implicit declaration of function 'jpeg_c_set_bool_param' [-Wimplicit-function-declaration]
             jpeg_c_set_bool_param(&cinfo, JBOOLEAN_TRELLIS_QUANT, FALSE);
             ^
src/util.c:158:49: error: 'JBOOLEAN_TRELLIS_QUANT_DC' undeclared (first use in this function)
         if (jpeg_c_bool_param_supported(&cinfo, JBOOLEAN_TRELLIS_QUANT_DC)) {
                                                 ^
src/util.c:167:49: error: 'JBOOLEAN_OPTIMIZE_SCANS' undeclared (first use in this function)
         if (jpeg_c_bool_param_supported(&cinfo, JBOOLEAN_OPTIMIZE_SCANS)) {
                                                 ^
make: *** [Makefile:57: src/util.o] Error 1

Wrote this command file for myself in order to run jpeg-recompress (in Windows) over directories with multiple jpeg files used in web developing.
It creates a context menu in right click over dirs to run recompress and jpeg-compare.

Thought to share it just in case someone finds it useful..
More info here: https://github.com/nikant/jpeg-recompress-win-context

Currently the makefile assumes lib64 is a part of the filesystem, on debian with multiarch and other multiarch distros lib64 doesn't exist. There are also other issues such as no DESTDIR and assuming mozjpeg is in /opt. I currently just sed -i '/^ifeq/,/^endif/d' and LIBJPEG=$PREFIX/lib/lib...

I see that there is a script called "butteraugli" that compare two images and give a score : https://github.com/google/butteraugli

I think that could be an interesting addition for jpeg-recompress.

I expect jpeg-recompress to read from stdin.

cat in.jpg | ./jpeg-recompress > out.jpeg

I use the date and time to remember when a picture was taken, it would be great if that that date/time was transferred to the new file as an option of jpeg-recompress.

According to wikipedia, this algorithm is a variation of SSIM that has following advantages

Three-component SSIM (3-SSIM) is a form of SSIM that takes into account the fact that the human eye can see differences more precisely on textured or edge regions than on smooth regions.[4] The resulting metric is calculated as a weighted average of SSIM for three categories of regions: edges, textures, and smooth regions. ...
This SSIM variant gives results which are more consistent with human subjective perception

Adding this would help in cases when preserving the image texture is important.

Thanks for this great tool! Would it be possible to add support for using TIFFs as the input? I'm a web developer at an art museum, working on automating image conversion for our website. It's pretty much the universal standard in the museum world to use TIFF as the archival format for object images. I could convert TIFFs to PPM and take it from there, but I'm running into some color balancing issues (which I'll fix shortly), and I feel like other folks out there might also appreciate the option to use TIFFs as a default, so that there's no need to bring in other utilities into the tool chain.

Cheers, and thanks again!

Edit: For the record, I fixed my issue with ImageMagick by extracting the color profile(s) from the TIFF before conversion to PPM, converting PPM to JPEG using jpeg-recompress, then applying the saved profile(s) to the JPEG. Here's the sequence of commands:

convert in.tif profile.icc
convert in.tif temp.ppm
jpeg-recompress --ppm temp.ppm out.jpg
convert out.jpg -profile profile.icc out-with-profile.jpg

I'll work on making this more efficient, but this is a valid work around for the moment.

´´´´
$ npm install imagemin-jpeg-recompress
npm WARN package.json [email protected] No repository field.
|

[email protected] postinstall C:\sites_testes\node_modules\imagemin-j
peg-recompress\node_modules\jpeg-recompress-bin
node lib/install.js

√ jpeg-recompress pre-build test passed successfully

[email protected] node_modules\imagemin-jpeg-recompress
├── [email protected]
├── [email protected] ([email protected], [email protected])
└── [email protected] ([email protected], [email protected])

C:>grunt imagemin
Running "imagemin:main" (imagemin) task
Warning: Error: Premature end of JPEG file
Bogus Huffman table definition
in file _images/mee.jpg Use --force to continue.

Aborted due to warnings.
´´´´

my task config is this:

´´´´
var imageCompressEngine = require('imagemin-jpeg-recompress');

var globalConfig = {

imagemin: {
  main: {
    options: {
      optimizationLevel: 7,
      pngquant: true,
      progressive: true,
      use: [imageCompressEngine()]
    },
    files: [
      {
        expand: true,
        cwd: '<%= cfg.root %><%= cfg.src.images %>',
        src: ['**/*.{png,jpg,jpeg,gif,ico}'],
        dest: '<%= cfg.root %><%= cfg.dest.images %>'
      }
    ]
  }
},

});
´´´´
My dependences are the latest versions of
"grunt-contrib-imagemin": "^0.9.2", and "imagemin-jpeg-recompress": "^4.0.0".

I'm working on windows OS, if i use another plugin, like imagemin-mozjpeg, it works perfectly.
I want to use this plugin, because it as a better compression.

Best regards, Mário Silva

@Lykegenes, it looks like you have a bunch of Windows-specific scripts in your fork at https://github.com/Lykegenes/jpeg-archive/tree/master/run - any chance you want to merge them into the official repo so that they can be part of the Windows release? Is it possible to get feature parity with the jpeg-archive bash script?

It would be really nice if I could pipe in and out of this utility.

I consider using jpeg-recompress at Fasterize. Fasterize automatically optimizes images from various websites.

So, images at the input may be uncompressed or already compressed at different levels.

1 - I wonder if there is a way to know if an image has already been compressed at a chroma subsampling 4:2:0 ?

2 - If my understanding is correct, jpeg-recompress picks the "right" quality score based on an image comparison. Is jpeg-recompress able to know which chroma subsampling is suitable for a given image ?
I think that if the input image is already with a subsampling 4:2:0, no more subsampling should be done.

Hi,

I enhanced your jpeg-archive script to be usable without any 3rd-party dependencies, and in-place compress either a single file or a whole folder (keeping all files meta-data).

You may want to include it in the project distribution (for which I of course grant you all rights):
https://gist.github.com/javabean/9ef827ea82268c1a9868ec617de3a1d8

We detected that jpeg images in the CYMK color system were not supported by jpeg-recompress.
However, mozjpeg seems to support this color system.

jpeg-recompress -Q --quality medium -m mpe -S disable ~/Desktop/3346470422995_patch.jpeg test.jpg
Unsupported color conversion request

I tried to use jpegoptim on the same image that also leverages on mozjpeg and jpegoptim correctly compresses the image.

I think that the right color system is not correctly detected by jpeg-recompress and thus jpeg-recompress doesn't correctly call mozjpeg methods.

The SSIMulacra metric promises significant Improvements over MS-SSIM. Among other things, it might finally solve the issue where images with a big, uniform background and a small, detailed object in the foreground get way too high scores, despite the foreground object being horribly garbled.

Any chance of this being implemented in jpeg-recompress?

http://cloudinary.com/blog/detecting_the_psychovisual_impact_of_compression_related_artifacts_using_ssimulacra

https://github.com/cloudinary/ssimulacra

Debian wheezy 64-bit, inside VMWare Workstation (64-bit host OS)

When building jpeg-archive it can't detect the bitness of the OS and it tries to load the 32 bit library instead:

cc: error: /opt/libmozjpeg/lib/libjpeg.a: No such file or directory

I edited the Makefile and changed the UNAME_P variable to call uname -m instead of uname -p. Then jpeg-archive built perfectly! 👍

I don't know about compatibility of this change in other systems, here's what I see:

# uname -a
Linux machinename 3.2.0-4-amd64 #1 SMP Debian 3.2.60-1+deb7u1 x86_64 GNU/Linux
# uname -m
x86_64
# uname -p
unknown

Currently jpeg-recompress segfaults when it cannot write to destination.

Hi!
Would it be possible to add support for JPG2K?
At least supporting also libjpeg alongside with mozjpeg.

Thank you!

Please consider allowing jpeg-compare to take two (instead of one) uncompressed PPM images.
This would make it much more flexible and for comparing other image compressors.

https://gist.github.com/sergejmueller/088dce028b6dd120a16e

Lossless and progressive optimization of multiple (random picked, less than 1 MB) JPEG images with mozjpeg 2.0 and libjpeg-turbo.

Both libraries are called with:

time jpegtran -copy none -progressive img.jpg > compressed.jpg

Summary

• mozjpeg generates 1,18% smaller* files as libjpeg-turbo
• mozjpeg is 2,3x slower* as libjpeg-turbo

* Average based on table entries

Whenever I run jpeg-archive, it puts everything into a Comp folder. Is there a way to have it replace the images it's recompressing?

Please add support for win32 (x86) for xp --> win10 as support seems broken since v2.1.0

Two test cases:

• jpeg-compare right_move_bg.jpg right_move_bg.jpg
Segmentation fault
  Download: right_move_bg.jpg

• jpeg-compare line.jpg line.jpg
*** Error in jpeg-compare: free(): invalid pointer: 0x8055b220 ***
======= Backtrace: =========
/lib/i386-linux-gnu/libc.so.6(+0x6929b)[0xb761429b]
/lib/i386-linux-gnu/libc.so.6(+0x6f527)[0xb761a527]
/lib/i386-linux-gnu/libc.so.6(+0x6fcd1)[0xb761acd1]
jpeg-compare(+0x293b)[0x8009693b]
jpeg-compare(+0x17f4)[0x800957f4]
jpeg-compare(main+0x14a)[0x8009555a]
/lib/i386-linux-gnu/libc.so.6(__libc_start_main+0xf7)[0xb75c3517]
jpeg-compare(+0x15a5)[0x800955a5]
Aborted
  Download: line.jpg

I need for some images to keep the color profile (ICC) in order to avoid color degradation. The original images is in AdobeRBG (rather than sRGB).

I see that jpegoptim has this kind of choice :

-s, --strip-all   strip all markers from output file
  --strip-none      do not strip any markers
  --strip-com       strip Comment markers from output file
  --strip-exif      strip Exif markers from output file
  --strip-iptc      strip IPTC/Photoshop (APP13) markers from output file
  --strip-icc       strip ICC profile markers from output file
  --strip-xmp       strip XMP markers markers from output file

Do you know if keeping the ICC profile is easily feasible in jpeg-recompress ?

Hi there,

first of all, thanks for that great work. We're considering jpeg-recompress to use in production, but we experience strange behavior when we have text in that image.

I've tried all different compression methods, but text gets blurry with all of them. See attached images. I even tried to use best quality and accurate settings:

jpeg-recompress --accurate --quality veryhigh --min 100 image.jpg image_recompress.jpg

You have any ideas what to do to to get this text sharp? (Adept.sh and jpeg-mini gets it sharp anyway)

Image before recompress:

After recompress:

Hey all,

I am on 32 bit Ubuntu. I have the latest mozjpeg and jpeg-archive. I copied the jpeg-archives binaries into /usr/local/bin/ and made sure there are +x.

However when I try jpeg-compress on an image it get: -bash: /usr/local/bin/jpeg-recompress: cannot execute binary file.

Is anyone else experiencing this?

More info here at the official repo and the blog.

i am getting error version `glibc_2.14' not found centos when i checked with yum it shows 2.12 .i tried to change version no in bash file but it is unreadable. can you please tell which release uses 2.12 glibs .i am using centos 6.8 and i checked the latest release through yum update and in their repo. its 2.12 only.
i can see only bash file that too without .sh extension. plz tell i have to make frm source or jpeg-archive-2.1.1-linux.tar.bz2
plz help

jpeg-archive works with mozjpeg 2.1 but not with 3.0 released a few days ago:

tpetry-macbook:jpeg-archive-2.0.1 tpetry$ make
cc -std=c99 -Wall -O3 -I/usr/local/Cellar/mozjpeg/include -c -o src/util.o src/util.c
src/util.c:128:15: error: no member named 'use_moz_defaults' in 'struct jpeg_compress_struct'
        cinfo.use_moz_defaults = TRUE;
        ~~~~~ ^
src/util.c:137:15: error: no member named 'optimize_scans' in 'struct jpeg_compress_struct'
        cinfo.optimize_scans = FALSE;
        ~~~~~ ^
2 errors generated.
make: *** [src/util.o] Error 1

The main problem is that homebrew has updated the mozjpeg formula to 3.0 and now you have to compile mozjpeg by your own to be able to compile jpeg-archive.

hi,

I have a problem with compressing images from standard input using jpeg-recompress.exe. The source files are different, so I have to use a program that opens various image files, eg. ffmpeg. + I use an additional program BATCHENC. The problem occurs when streaming ppm to jpeg-recompress.

example

ffmpeg.exe -i <infile> -f image2pipe -vcodec ppm - | jpeg-recompress.exe --ppm - out.jpg

there is an error

Incorrect image size! 374 vs. 5992721
av_interleaved_write_frame(): Broken pipe
frame=    1 fps=0.0 q=0.0 Lsize=    5852kB time=00:00:00.04 bitrate=1198544.2kbits/s
video:5852kB audio:0kB subtitle:0kB other streams:0kB global headers:0kB muxing overhead: 0.000000%
Conversion failed!

example 2
ImageMagick also does not work

convert.exe <infile> ppm:- | jpeg-recompress.exe --method smallfry --ppm - out.jpg

there is an error

Incorrect image size! 491 vs. 5992721

Am I doing something wrong?

Hi,
I was unable to find a ladon or a jpeg-archive windows executable. Could you please kindly add the instructions so that I can execute your very promising jpeg-recompress on multiple recursive JPG in windows?
I saw a while a ago one of your users proposed a branch to allow recursive processing directly in jpeg-recompress: was this never adopted?
thanks and keep up the good work!

I am trying to build jpeg-archive on Mac OS X Yosemite 10.10.2, and it is throwing me the following error.

cc -std=c99 -Wall -O3 -I/opt/mozjpeg/include -c -o src/util.o src/util.c
src/util.c:133:13: warning: implicit declaration of function 'jpeg_c_bool_param_supported' is invalid in C99 [-Wimplicit-function-declaration]
        if (jpeg_c_bool_param_supported(&cinfo, JBOOLEAN_OPTIMIZE_SCANS)) {
            ^
src/util.c:133:49: error: use of undeclared identifier 'JBOOLEAN_OPTIMIZE_SCANS'
        if (jpeg_c_bool_param_supported(&cinfo, JBOOLEAN_OPTIMIZE_SCANS)) {
                                                ^
src/util.c:134:13: warning: implicit declaration of function 'jpeg_c_set_bool_param' is invalid in C99 [-Wimplicit-function-declaration]
            jpeg_c_set_bool_param(&cinfo, JBOOLEAN_OPTIMIZE_SCANS, FALSE);
            ^
src/util.c:134:43: error: use of undeclared identifier 'JBOOLEAN_OPTIMIZE_SCANS'
            jpeg_c_set_bool_param(&cinfo, JBOOLEAN_OPTIMIZE_SCANS, FALSE);
                                          ^
2 warnings and 2 errors generated.
make: *** [src/util.o] Error 1

Any ideas, may I missing any dependency ?

Sorry, this isn't a bug per se, just a heads-up to anyone else who might encounter this issue. If the image you are trying to convert is more than 8 bits per channel, jpeg-recompress will throw an error: Unsupported bit depth #####!.

This is definitely something to watch out for when you are batch-converting with a custom script. I tried converting PPMs generated from 16 bpc TIFFs, and the image numbers just didn't match up. Took me a couple days to realize what was going on.

You can solve this problem with ImageMagick by specifying a depth parameter:

convert in.tif profile.icc
convert in.tif -depth 8 temp.ppm
jpeg-recompress --ppm temp.ppm out.jpg
convert out.jpg -profile profile.icc out-with-profile.jpg

Additionally, it seems that TIFFs might store their bit depth as a floating point value, not an integer, which is why the error I saw read out as Unsupported bit depth 65535!. I'm not completely certain this is the case, just a hunch. Cheers!