[Matroska-devel] JPEG 2000 codec support

Antonin Descampe antonin.descampe at uclouvain.be
Tue Feb 17 14:01:56 CET 2015


@Moritz and Alexander: thanks for the info and links.

@wm4: a few comments hereunder.

Le 13 févr. 2015 à 16:14, wm4 <nfxjfg at googlemail.com<mailto:nfxjfg at googlemail.com>> a écrit :

On Fri, 13 Feb 2015 14:58:30 +0000
Antonin Descampe <antonin.descampe at uclouvain.be<mailto:antonin.descampe at uclouvain.be>> wrote:


Le 13 févr. 2015 à 12:06, wm4 <nfxjfg at googlemail.com<mailto:nfxjfg at googlemail.com>> a écrit :

On Wed, 4 Feb 2015 15:37:29 +0000
Antonin Descampe <antonin.descampe at uclouvain.be<mailto:antonin.descampe at uclouvain.be>> wrote:

Dear Matroska developers ,

In the context of an EU-funded project called PREFORMA dealing with long-term preservation archives, we need to wrap JPEG 2000 frames using MKV.

That sounds insane to me. Why did they choose JPEG 2000? It has bad
support in open source and AFAIR has nothing on h264 as far as
compression efficiency and quality goes.

Well, I have quite a different opinion.
Concerning open-source support, there is the openjpeg library.
Concerning compression efficiency and quality, JPEG 2000 and H264 are suited for completely different use cases. In the case of long-term preservation  archives where they need high bitrate (possibly lossless) compression, JPEG 2000 makes perfectly sense. That’s probably why well-known institutions like INA (french archives), British library, or Library of congress have opted for J2K for their assets. In addition to this, J2K is royalty-free, which is not the case of h264.

OpenJPEG is slow (and the encoder is low quality, if that matters
here).

OpenJPEG is slow indeed. Wider support from the open-source community (through synergies with the above-mentioned EU project for instance) would surely improve this. I would actually be glad if this happens as I’m one of the OpenJPEG maintainers. Concerning the encoder being low quality, I respectfully disagree with you, although I’m aware a lot of improvements can still be made.

JPEG 2000 still has a patent problem.

Yes … The JPEG 2000 patent problem. Well, long story short, IMHO there is more a video and image compression patent problem than a JPEG 2000 patent problem. Image and video compression are heavily patented fields. JPEG Committee knows this perfectly well and always develops its standards by asking companies to state patents they might have and negotiates with them free or RAND licenses in the case they would like to exploit their patent (which has never happen for J2K although it is used in high-value markets such as digital cinema). If such licenses cannot be obtained from the companies, the patented technology is removed from the standard.  Note that JPEG and MPEG do not deal with this issue the same way, as up to now JPEG always released royalty-free standards, which is not the case of MPEG. Anyway, I think this is actually a safer path than developing a compression codec outside of standardization organizations and claiming it’s open and unpatented. By the way, technologies used in JPEG 2000 Part-1 are all almost twenty years old now.

JPEG 2000 in Matroska is
so obscure that you're probably the first person asking for it - how do
you expect the future to deal with these files?

I do not see anything obscure here and I cannot foresee the future. Anyway, J2K wrapped in MKV is a requirement from the EU project. Personally I would have chosen MXF instead of MKV as MXF is the usual wrapper for J2K in broadcast and archiving. But I’m happy to dig in MKV and contribute to it if needed.


Why not use ffv1? Lossless, lower size and faster than jpeg 2000,
broader support than jpeg 2000, no patents, supported in Matroska, and
apparently also used for archiving.

No doubt FFV1 is a very good codec. And if lossless compression is the only requirement for long-term preservation, it’s surely a good candidate. JPEG 2000 has much more features to offer, mainly in terms of scalability and accessibility, and that’s also why it’s more complex. So the question is more on the long term preservation requirements.


So, my question is: why not choose something that has good open source
support, is widely used, isn't horribly complex and obscure (think of
the future), and has better lossless compression with smaller size and
higher performance?

The reason that JPEG 2000 is pushed is likely not on technical merits,
but because big companies want to get some of that tax payer's money.

Something complex is not necessarily obscure :-). I’m convinced JPEG 2000 is still one of the most advanced and future-proof still-image compression standard out there. Its main drawback (beside its tacky name of course) is actually that it has so many technical merits that it’s probably a bit complex, which slowed down its adoption, in particular in the open source community. Companies are not pushing JPEG 2000 for bad evil reasons. They are looking for efficient, full-featured, royalty-free, interoperable compression formats, and JPEG 2000, as an ISO standard, just gives them what they want.

Cheers,

Antonin




I looked in mail archives and documentation, but it seems there is currently no such JPEG 2000 support within Matroska.

My questions:
* Is there any plan to include JPEG 2000 as one of the supported codec in Matroska ? If yes, what would be the timeframe ?
* Is there any work already done in that direction ?
* What would be the steps to follow to get this codec supported by Matroska ?

Many thanks in advance for any comment on that topic !

Cheers,

Antonin

_______________________________________________
Matroska-devel mailing list
Matroska-devel at lists.matroska.org<mailto:Matroska-devel at lists.matroska.org>
http://lists.matroska.org/cgi-bin/mailman/listinfo/matroska-devel
Read Matroska-Devel on GMane: http://dir.gmane.org/gmane.comp.multimedia.matroska.devel

_______________________________________________
Matroska-devel mailing list
Matroska-devel at lists.matroska.org<mailto:Matroska-devel at lists.matroska.org>
http://lists.matroska.org/cgi-bin/mailman/listinfo/matroska-devel
Read Matroska-Devel on GMane: http://dir.gmane.org/gmane.comp.multimedia.matroska.devel
_______________________________________________
Matroska-devel mailing list
Matroska-devel at lists.matroska.org<mailto:Matroska-devel at lists.matroska.org>
http://lists.matroska.org/cgi-bin/mailman/listinfo/matroska-devel
Read Matroska-Devel on GMane: http://dir.gmane.org/gmane.comp.multimedia.matroska.devel

-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.matroska.org/pipermail/matroska-devel/attachments/20150217/e1155417/attachment-0001.html>


More information about the Matroska-devel mailing list