Menu
▾
▴
jpeg-users
|
From: Bill A. <Bil...@ma...> - 2008-11-26 20:05:42
|
Hello Guido, It has been brought to my attention: your post from 2006 http://groups.google.com/group/comp.compression/msg/77d99c47ec1736da?hl=en& and the file http://jpegclub.org/libjpeg-6c.tar.gz Is it an official libjpeg release ? Should I upgrade the Debian package to it ? What are the changes ? Cheers, Bill. |
|
From: Bill A. <Bil...@ma...> - 2008-11-30 18:50:10
|
On Wed, Nov 26, 2008 at 09:05:33PM +0100, Bill Allombert wrote: > Hello Guido, > > It has been brought to my attention: your post from 2006 > http://groups.google.com/group/comp.compression/msg/77d99c47ec1736da?hl=en& > and the file http://jpegclub.org/libjpeg-6c.tar.gz > > Is it an official libjpeg release ? Guido answered that this is not an official release, and that I should wait for v7. Cheers, Bill. |
|
From: Mathieu M. <mat...@gm...> - 2008-11-30 18:58:48
|
On Sun, Nov 30, 2008 at 7:50 PM, Bill Allombert <Bil...@ma...> wrote: > On Wed, Nov 26, 2008 at 09:05:33PM +0100, Bill Allombert wrote: >> Hello Guido, >> >> It has been brought to my attention: your post from 2006 >> http://groups.google.com/group/comp.compression/msg/77d99c47ec1736da?hl=en& >> and the file http://jpegclub.org/libjpeg-6c.tar.gz >> >> Is it an official libjpeg release ? > > Guido answered that this is not an official release, and that I should > wait for v7. Cool ! Thanks for the info. No roadmap, no timeline ? Is libjpeg6c stable ? Where does the development take place ? Thanks, -- Mathieu |
|
From: Mathieu M. <mat...@gm...> - 2008-12-02 10:02:30
|
Fwding as Guido is not member of jpeg-users ---------- Forwarded message ---------- From: Guido Vollbeding <gu...@jp...> Date: Sun, Nov 30, 2008 at 10:04 PM Subject: Re: [Jpeg-users] libjpeg 6c ? To: Mathieu Malaterre <mat...@gm...> Cc: Bill Allombert <Bil...@ma...>, jpe...@li... Hello all Mathieu Malaterre wrote: > > Cool ! Thanks for the info. > No roadmap, no timeline ? I have a roadmap, but no timeline. > Is libjpeg6c stable ? It appears to be stable, as far as I can tell. But I'm actually not in the condition to conduct the necessary tests for all the platforms, and some minor things are still missing (particularly updated documentation), so that's why it is not a release, although the major new features for the upcoming v7 update are already contained in this package. > Where does the development take place ? The development is actually exclusively in my hands. Best regards Guido Vollbeding Organizer Independent JPEG Group -- Mathieu |
|
From: paul v. d. b. <pau...@gm...> - 2008-12-02 19:50:52
|
On Wed, 26 Nov 2008 21:05:33 +0100 Bill Allombert <Bil...@ma...> wrote: > Is it an official libjpeg release ? > Should I upgrade the Debian package to it ? > What are the changes ? I have followed http://jpegclub.org/libjpeg-6c.tar.gz for some time. It is a moving target. I have at least 4 different versions, 2 from 2006 and 2 from 2008. The most important changes from the original ijg-release are: cjpeg -h adds: -scale M/N Scale image by fraction M/N, eg, 1/2 djpeg -h adds: -scale M/N Scale output image by fraction M/N, eg, 1/8 jpegtran -h adds: -crop WxH+X+Y Crop to a rectangular subarea [ this is already included in debian, gentoo, suse ..? ] But if you run 'make -i test' on the 6c source with testimages from ijg-source you get: ./djpeg -dct int -ppm -outfile testout.ppm ./testorig.jpg ./djpeg -dct int -bmp -colors 256 -outfile testout.bmp ./testorig.jpg ./cjpeg -dct int -outfile testout.jpg ./testimg.ppm ./djpeg -dct int -ppm -outfile testoutp.ppm ./testprog.jpg ./cjpeg -dct int -progressive -opt -outfile testoutp.jpg ./testimg.ppm ./jpegtran -outfile testoutt.jpg ./testprog.jpg cmp ./testimg.ppm testout.ppm ./testimg.ppm testout.ppm differ: byte 37, line 4 make: [test] Error 1 (ignored) cmp ./testimg.bmp testout.bmp ./testimg.bmp testout.bmp differ: byte 59, line 1 make: [test] Error 1 (ignored) cmp ./testimg.jpg testout.jpg ./testimg.jpg testout.jpg differ: byte 1395, line 7 make: [test] Error 1 (ignored) cmp ./testimg.ppm testoutp.ppm ./testimg.ppm testoutp.ppm differ: byte 37, line 4 make: [test] Error 1 (ignored) cmp ./testimgp.jpg testoutp.jpg ./testimgp.jpg testoutp.jpg differ: byte 1582, line 8 make: [test] Error 1 (ignored) cmp ./testorig.jpg testoutt.jpg Regards, Paul |
|
From: Mathieu M. <mat...@gm...> - 2008-12-02 20:01:19
|
On Tue, Dec 2, 2008 at 8:50 PM, paul van den berg <pau...@gm...> wrote: > On Wed, 26 Nov 2008 21:05:33 +0100 > Bill Allombert <Bil...@ma...> wrote: > >> Is it an official libjpeg release ? >> Should I upgrade the Debian package to it ? >> What are the changes ? > > I have followed http://jpegclub.org/libjpeg-6c.tar.gz for some time. > It is a moving target. I have at least 4 different versions, 2 from 2006 > and 2 from 2008. > > The most important changes from the original ijg-release are: > > cjpeg -h adds: > -scale M/N Scale image by fraction M/N, eg, 1/2 > > djpeg -h adds: > -scale M/N Scale output image by fraction M/N, eg, 1/8 > > jpegtran -h adds: > -crop WxH+X+Y Crop to a rectangular subarea > [ this is already included in debian, gentoo, suse ..? ] > > But if you run 'make -i test' on the 6c source with testimages > from ijg-source you get: > > ./djpeg -dct int -ppm -outfile testout.ppm ./testorig.jpg > ./djpeg -dct int -bmp -colors 256 -outfile testout.bmp ./testorig.jpg > ./cjpeg -dct int -outfile testout.jpg ./testimg.ppm > ./djpeg -dct int -ppm -outfile testoutp.ppm ./testprog.jpg > ./cjpeg -dct int -progressive -opt -outfile testoutp.jpg ./testimg.ppm > ./jpegtran -outfile testoutt.jpg ./testprog.jpg > cmp ./testimg.ppm testout.ppm > ./testimg.ppm testout.ppm differ: byte 37, line 4 > make: [test] Error 1 (ignored) > cmp ./testimg.bmp testout.bmp > ./testimg.bmp testout.bmp differ: byte 59, line 1 > make: [test] Error 1 (ignored) > cmp ./testimg.jpg testout.jpg > ./testimg.jpg testout.jpg differ: byte 1395, line 7 > make: [test] Error 1 (ignored) > cmp ./testimg.ppm testoutp.ppm > ./testimg.ppm testoutp.ppm differ: byte 37, line 4 > make: [test] Error 1 (ignored) > cmp ./testimgp.jpg testoutp.jpg > ./testimgp.jpg testoutp.jpg differ: byte 1582, line 8 > make: [test] Error 1 (ignored) > cmp ./testorig.jpg testoutt.jpg > That's pure gold ! :) Thanks I added that info to jpeg's wiki: http://apps.sourceforge.net/mediawiki/jpeg/index.php?title=Jpeg-6c Thanks -- Mathieu |
|
From: Guido V. <gu...@jp...> - 2008-12-02 21:25:35
|
paul van den berg wrote: > > But if you run 'make -i test' on the 6c source with testimages > from ijg-source you get: But why would you run the test of 6c with testimages from 6b? One of the improvements of this 6c package is that it has got its own corrected testimages! The testimages from 6b cannot work with 6c/7, because I have introduced a more correct method for the internal color subsampling processing. The method used in 6b is inferior and has serious shortcomings. 6c/7 solves these problems by using the same new direct DCT scaling functions for the color subsampling as for the image rescaling. The compressed testimages had to be adapted (recreated with new method) for the new version. There is nothing wrong with this, it's actually a feature, not a bug! Regards Guido Vollbeding Organizer Independent JPEG Group |
|
From: Mathieu M. <mat...@gm...> - 2008-12-02 21:43:19
|
On Tue, Dec 2, 2008 at 10:20 PM, Guido Vollbeding <gu...@jp...> wrote: > paul van den berg wrote: >> >> But if you run 'make -i test' on the 6c source with testimages >> from ijg-source you get: > > But why would you run the test of 6c with testimages from 6b? > One of the improvements of this 6c package is that it has got > its own corrected testimages! The testimages from 6b cannot > work with 6c/7, because I have introduced a more correct method > for the internal color subsampling processing. The method used > in 6b is inferior and has serious shortcomings. 6c/7 solves > these problems by using the same new direct DCT scaling > functions for the color subsampling as for the image rescaling. > The compressed testimages had to be adapted (recreated with new > method) for the new version. There is nothing wrong with this, > it's actually a feature, not a bug! I updated the wiki. Thank you Guido. http://apps.sourceforge.net/mediawiki/jpeg/index.php?title=Jpeg-6c&curid=9&diff=40&oldid=39&rcid=45 -- Mathieu |
|
From: Guido V. <gu...@jp...> - 2008-12-02 22:41:34
|
By the way, you may consider this one of the reasons why I will not provide any official v6 update anymore. The next release will be v7, because there are fundamental enhancements in this version. Regards Guido Vollbeding Organizer Independent JPEG Group |
|
From: Mathieu M. <mat...@gm...> - 2008-12-02 22:57:05
|
On Tue, Dec 2, 2008 at 11:41 PM, Guido Vollbeding <gu...@jp...> wrote: > By the way, you may consider this one of the reasons why I > will not provide any official v6 update anymore. > The next release will be v7, because there are fundamental > enhancements in this version. ok. By the way do you consider merging the lossless patch from Ken Murchison in v7 at all ? I have tried, but I really lack an in depth understanding of IJG (and jpeg spec) to be able to finish the work. Thank you. -- Mathieu |
|
From: Guido V. <gu...@jp...> - 2008-12-02 23:38:41
|
Hi Mathieu The most important aspect of that v7 update will be to introduce new features in the core of the JPEG system, and that core is the DCT! This has to be understood first: The primary aspect is to utilize the fundamental DCT properties to a more appropriate degree than has been done before in image coding in general, and everything else is secondary! You must understand that all established image compression research and development is actually going in wrong directions, and the reason for this situation is that the DCT and its fundamental properties are not understood properly. There is a common lack of understanding the basic image coding principles - I have until today not come across a single person who would understand the essence of the DCT for image coding! And I can tell you that I have met and communicated with lots of so-called "experts" in this field - including the original JPEG and MPEG authors! And no one of them has understood the DCT and its potential! So that is the primary and essential goal to achieve. Without proper understanding and utilizing the DCT there can be no real progress in image coding - all other attempts which we see today are mistakes and will fail, due to misunderstanding of and deviating from the true DCT. JPEG as we know it today uses only a tiny subset of the full DCT potential, and v7 starts to extend the features to a more appropriate degree. (It lays foundation for other important features to come later...) > By the way do you consider merging the lossless patch from Ken > Murchison in v7 at all ? You mean the lossless mode of the original JPEG spec? There is a reason that it has not found noticeable application in practice. It is not seamlessly integrated with the usual DCT processing mode. With proper understanding of the DCT you can provide a lossless mode which is seamlessly integrated with the DCT mode. This is one of the features possible with the new direct DCT scaling functions as described on my site and in my JPEG-Plus proposal, which is actually ignored at the official committees because they are invested in bogus developments. Regards Guido Vollbeding Organizer Independent JPEG Group |
|
From: paul v. d. b. <pau...@gm...> - 2008-12-03 09:58:04
|
On Wed, 03 Dec 2008 00:38:47 +0100 Guido Vollbeding <gu...@jp...> wrote: > ... This is one of the features possible with the > new direct DCT scaling functions as described on my site and in > my JPEG-Plus proposal ... Can you give any information on your JPEG-Plus proposal? > Regards > Guido Vollbeding > Organizer Independent JPEG Group What group are you organizing? Who is involved? Can we track its progress? Regards, Paul |
|
From: Mathieu M. <mat...@gm...> - 2008-12-03 10:22:26
|
On Wed, Dec 3, 2008 at 10:57 AM, paul van den berg <pau...@gm...> wrote: > On Wed, 03 Dec 2008 00:38:47 +0100 > Guido Vollbeding <gu...@jp...> wrote: > >> ... This is one of the features possible with the >> new direct DCT scaling functions as described on my site and in >> my JPEG-Plus proposal ... > > Can you give any information on your JPEG-Plus proposal? > After a quick google search: http://jpegclub.org/temp/ITU-T-JPEG-Plus-Proposal.doc 2cts -- Mathieu |
|
From: Guido V. <gu...@jp...> - 2008-12-03 10:57:14
|
paul van den berg schrieb: > > Can you give any information on your JPEG-Plus proposal? I'm really puzzled now. Is this the same "paul van den berg" with whom I had a discussion about this subject in 2006 when this document was created? Also, the "paul van den berg" back then provided a pdf conversion of this document. It is all available here: http://jpegclub.org/temp/ http://jpegclub.org/temp/ITU-T-JPEG-Plus-Proposal_R3.doc There have been more things happening in the meantime and other documents being written (which are not publicly available) with more descriptions and more discoveries described, but this one is still valid. > What group are you organizing? > Who is involved? Can we track its progress? It is the entity which I took over from the previous IJG organizer Tom Lane. The "group" was nothing more than a mailing list, and unfortunately it went defunct some time back. Then I tried to recover the list by taking it over to my own site, but it turned out that the effort wasn't worth it because there were no more substantial contributions. And I think the group was not very helpful even before that happened - the IJG libjpeg effort was basically the whole work of Tom Lane himself, and now it is basically the whole work of myself. There is no noticeable support from anywhere, that's why Tom Lane couldn't continue the work. I took it over because I made fundamental discoveries and contributions, and nobody else was present to do that. The missing support problem is still the same, but what can I do? Either I do it, or nobody else will do it, and DCT-based JPEG, which is the only proper base for image coding application, would die. Yes, there are always many people demanding support, updates, and asking questions, and I do my best to answer the questions and hint to solutions, but there is no one who would provide any fund for doing this work. There are many funds available, but, as said, they are actually all invested in bogus developments. This has to be understood. And as long as this situation continues, there can be no proper progress. What I'm actually doing is to establish an enterprise for realizing the discovered features in the movie coding domain! There are much more opportunities available in the movie coding. The still image domain is actually in complete darkness - all research, development, and industry is corrupted in this area. I have now found some kind of support for introducing the features for movie coding applications, and if this goes well, I will also be able to push the still image develeopment in the right direction. Regards Guido Vollbeding Organizer Independent JPEG Group |
|
From: paul v. d. b. <pau...@gm...> - 2008-12-03 20:25:48
|
On Wed, 03 Dec 2008 11:55:41 +0100 Guido Vollbeding <gu...@jp...> wrote: > paul van den berg schrieb: > > > > Can you give any information on your JPEG-Plus proposal? > > I'm really puzzled now. > Is this the same "paul van den berg" with whom I had a discussion about this > subject in 2006 when this document was created? Yes, I am still the same. > Also, the "paul van den berg" back then provided a pdf conversion of this document. > It is all available here: > > http://jpegclub.org/temp/ > http://jpegclub.org/temp/ITU-T-JPEG-Plus-Proposal_R3.doc I am glad that you make this resource available to this list. I privately knew that it existed but it is up to you to make it public. Even your own site does not refer to this document. _R3 is more recent than the link Mathieu gave. A Google-search on 'jpeg-plus' also gives this link: www.itu.int/md/T05-SG16-060403-D-0287/en but the document it refers to is not publicly available. Could you summarize for this list what happened? > > What group are you organizing? > > Who is involved? Can we track its progress? [...] Thanks for sharing these insights. I am sure that very substantial parts in IJG-libjpeg are not coded by Tom Lane. (See your own contributions). I would like to see an open future development of IJG-libjpeg. Regards, Paul |
|
From: Guido V. <gu...@jp...> - 2008-12-04 18:49:00
|
paul van den berg wrote: > > Could you summarize for this list what happened? This is a long story. The short story is that I'm now realizing the features in a movie production environment, based on a Motion-JPEG codec. It turns out that the SmartScale option is the key feature, that's why the current working name is "SmartScale Motion Picture Codec". It also turns out that for HD (High Definition) movie processing it is not sufficient to implement the core functions in C code - you need to optimize the functions on assembly code basis with extended media instruction sets (MMX, SSE...) for realtime application in high definition movie processing. That's what I'm actually busy with, and that's why I can't tell you the long story now... > I would like to see an open future development of IJG-libjpeg. An open development has not helped in the past - as said the IJG/ libjpeg development was basically an individual effort. I'm actually quite confident that my current involvement in the movie coding field will provide a well foundation for coming further in the right direction. It considerably extends the practical experiences, and, hopefully, can generate some fund for other work... Regards Guido |
Thanks for helping keep SourceForge clean.
X