VP3.com: Future VP3 Releases To Be LGPL

sudog writes: "According to this vorbis-dev posting and The VP3 Homepage VP3 (QT5-type movie compression scheme) is now under the LGPL! What's not clear is whether they intend to offer it guaranteed royalty and patent free to the community. They're actively looking for help, too. Does this mean that we no longer need the OGG-Tarkin to save us from our movie-less, video-app-less emulating?" Of course, they don't say starting when, exactly.

Comments

QT is a metaformat..

[snillfisk]

After all, QT is a metaformat .. which isnt actually doing any compression or anything, but rather specifying which codecs to use. the biggest problem with getting an open source version that handles all QT-files nicely, is the patents and licensing on central codecs (especially the Sorenson Video codec).. more info can be found at http://www.apple.com/quicktime/specifications.html .

Patent protection - from Vorbis mailing list

[SpinyNorman]

___ Dan Miller
(++,) CTO and founder, On2 Technologies

On Wed, 27 Mar 2002 volsung@asu.edu wrote:

> On Wed, 27 Mar 2002, Daniel B. Miller wrote:
>
> > Hi fellow Ogg-oids --
> >
> > I wanted to let everyone know that VP3, our open-source video codec that
> > is commonly used with QT5, is being re-released under the LGPL. We are
>
> Really!?! All I can say is wow. What about the patent issues? Are you
> granting royalty-free license to the required patents along with the license
> to the code?

That's a requirement of the LGPL, so, yes.

Re:Patent protection - from Vorbis mailing list

[AVee]

Hey, stop it, that was my post. ;-)
Well I'll post the link then...

Re:Patent protection - from Vorbis mailing list

[Rentar]

Wow! I defintely think this is basically a good thing, but I got two objections (or rather questions, or ideas that got something to do with it, or ... whatever):

• How do they stay in buisness? Re-licensing it for money? Support? If so: Good luck, you'll need it!
• I hope this won't prevent the tarkin-folks from trying to come up with their own new-and-improved codec.

Re:Patent protection - from Vorbis mailing list

[AVee]

How do they stay in buisness? Re-licensing it for money?
Read this: The power of this codec has been validated by the fact that it has been licensed by both RealNetworks and Apple for their internet video players.

They do license it right now. They might get in to trouble with this though, since they can't forbid anybody to use the LGPL'ed version or make them pay for it once it is released to the public. So they might get a lot of support from the OSS community, wich will improve their software, but likely not their income...

Re:Patent protection - from Vorbis mailing list

[Anonymous Coward]

VP3 isn't their newest codec

so you get
1. the lgpl'd version or
2. the closed sourced, expensive licensed but more powerful new version

Speaking of quicktime...

[xtermz]

...The latest issue of Linux Journal (remember that mag? ) has a good article on the various quicktime players available for Linux, and how to install/use them ...

Check them here : Linux Journal

Re:Speaking of quicktime...

[Junta]

Of course, the problem here is that the native players don't support Sorenson, and since that by and large is the format of quicktime on the interne,t it is a moot point. I'll stick with wine..

I'd never trust anyone except Ogg

[raahul_da_man]

Another video codec sounds good, but after the nasty DivX ;) experience, I don't trust people who make their codecs open source, and then return to being proprietary.

LGPL is a bastard license at best. GPL or nothing is the only way to go.

Re:I'd never trust anyone except Ogg

[JabberWokky]

DivX ;) has had a very bumpy and very forked road - primarily because it wasn't a file format so much as a hacked DLL that was a stolen Microsoft test codec, with the audio replaced with (originally) MP3. Since there was no formal definition of the file format, it's fragmented pretty heavily with different encoders producing output that works on some of the variants, and other encoders working on different variants. At least two efforts have been made to rewrite the whole thing from scratch, and at least two alternate sound codecs exist (thus the common cry of "there's no sound in that file!"). There are questions of patents being filed even now (since it was a "format" swiped from a corporation), and, all in all, the situation has been very tangled for a single, real standard to emerge.

LGPL grants the same public use that the GPL does, except you can also combine it with commercial software (you have to release source only for the modifications done to that specific part of the code). It's not a "bastard license", but rather a compromise to allow commercial software to link to fundimental system libraries and run on a Free system. YMMV on what you think of that, but for things like file formats and reference code to file formats, IMO, the LGPL is the best license around - it keeps the whole thing open, including any changes that anybody makes to it, so the standard is open for the whole world no matter who uses it in any application.

I'm personally of the opinion that an LGPL library to read a few types of XML documents (a word processor format, a spreadsheet format, a bitmap with annotations, a vector art format and a vector engineering format) should be made, and maintained by all major office suites, probably starting off with the various open source projects. Even if a company didn't use the exact code, it serves as reference code for compatability tests and extensions.

--
Evan

Re:I'd never trust anyone except Ogg

[sab39]

Do you see the contradiction in saying you never trust anyone except Ogg, and then insisting on GPL only? Hint: Ogg was originally LGPL, and is now X11 - even *less* GPL (and in Ogg's case, RMS agreed this was a good move, because of the network effect needed to get good Ogg adoption).

I was never really involved in the DivX;) scene, but by my understanding the DivX license was never *really* open source. If it had been, they *couldn't* "take it proprietary", just take *future versions* proprietary (in which case a fork would take place and other interested coders could keep the open source version going). Since this didn't happen, I'm assuming that the license wasn't really open source.

It should be noted that VP3 has been under an "allegedly" open source license for ages, but it was one of these SCSL- or BitKeeper-style things that requires you only use the code for developing codecs compatible with the VP3 spec. This made it not truly open source, and left me at least very unenthusiastic about it (at that point they *could* have made it proprietary, effectively, through messing with the "official" specs often enough to make following them impossible). Now it's under a truly free license (or at least once such a release is made) there's no reason to be dubious anymore.

Stuart.

Re:I'd never trust anyone except Ogg

[benwaggoner]

VP5 is a very different beast than VP3 - there was a whole VP4 version between there. I doubt On2 is getting much in the way of benefit to VP5 from VP3 being open source. It's probably mainly a brand recognition thing to keep their name in the codec community while they prepare VP5 for launch.

Re:I'd never trust anyone except Ogg

[Dwonis]

Sigh. The open-source community can always fork VP3 and compete with VP5 if they do that. That's the nature of the *GPL.

Future?

[O2n]

Future versions of VP3 will be released under the LGPL [...] Stay tuned!

In other news, future versions of Windows will be released under the LGPL; stay tuned and don't forget to tell your children to tell their children to tell...

:)

Re:Future?

[FreeMath]

According to the VP3 homepage the source is already available.

Re:Future?

[volsung]

The source has been available for a long time, but the source license has been non-free (and had some restrictions I personally considered unacceptable). The news here is that the source is now LGPL as opposed to the weird license that made it unusable before.

Looks like a good thing to me.

[noser]

I have never heard of this codec, but it seems to me that this is more or less what the LGPL is intended for. Take a quick look at the LGPL and note this section:

For example, on rare occasions, there may be a special need to encourage the widest possible use of a certain library, so that it becomes a de-facto standard. To achieve this, non-free programs must be allowed to use the library. A more frequent case is that a free library does the same job as widely used non-free libraries. In this case, there is little to gain by limiting the free library to free software only, so we use the Lesser General Public License.

(Emphasis mine)

Seems to me that the people at VP3 would like as many people as possible to start working with their codec, allowing it to gain ascendancy over other codecs so that someday they will be able to make money selling their own "enhanced" version. Not a bad deal for GNU, because we get something badly needed. I hope that we start to hear more about this codec being used in some interesting projects in the future now that it has become more available.

What's with XviD?

[nutshell42]

http://www.xvid.org

XviD is based on the old OpenDivX-Codec but afaik doesn't use any of its code any longer and is completely GPled.

the codec improves at an amazing speed and already beats the shit out of VP3:
http://www.doom9.org/codecs.htm

Re:What's with XviD?

[rsborg]

XviD falls prey to the problems inherent with MPEG4, as previously discussed here. Fix that, and I'm all up for it. (personally, I don't see any way around the issues)

In the interim, there's Ogg Tarkin, but it looks like they're too busy with Vorbis right now.

XVID, a GPL video codec

[BoBInO]

For those bitter about Divx, have a look at XVID (www.xvid.org). This a GPL video codec based on the source of OpenDivx.
This codec give me excellent results compared to Divx 3.11Alpha and Divx5...

With XVID video and OGG sound all in a OGM file (OGg Media) i get fully legal DVD-Rip !!! :)

Re:XVID, a GPL video codec

[Skuto]

>With XVID video and OGG sound all in a OGM file
>(OGg Media) i get fully legal DVD-Rip !!! :)

That's assuming XviD doesn't rely on any external MPEG4 patents, and as far as I know, it does.

It's in a similar situation as LAME. The code is GPL, but not legal to use in most countries due to patents. This is why the binaries are usually found on a Russian or Brazialian server.

--
GCP

Excellent move

[AirLace]

VP3 is a really powerful codec. It's like a next-generation DivX -- similarly to DivX the only thing holding back from widespread use were its worrying licensing terms. Now that it's free I'm looking forward to seeing this adopted fully by the community.

This means that streaming video embedded in webpages for the Linux / Free Software clan is now a reality, but perhaps more importantly, VP3 makes it possible to eliminate adware/spyware like RealVideo and the equally obnoxious and platform-specific Windows Media. Cheers for all the great work, On2 Technologies! This is, in my opinion, one of the most important things to happen on the open source scene for quite some time.

This doesn't obsolete Tarkin BTW :)

[Skuto]

Tarkin is currently working on bringing new technologies such as wavelets and 3-d transforms into video coding. It's not finished yet, but it offers more possibilities for really new technology and further development.

While this is great news, it by no means means that Ogg Tarkin suddenly is obsoleted :) VP3 is available now though, and Tarkin isn't.

--
GCP

A couple of points clarified

[AirLace]

• But the code is still patented? No! A post from Dan Miller, CEO of On2 Technologies on vorbis-dev:
  
  > Really!?! All I can say is wow. What about the patent issues? Are you
  > granting royalty-free license to the required patents along with the license
  > to the code?
  That's a requirement of the LGPL, so, yes
  
  
• But what about VP4? It's a better codec, and it's proprietary. VP3 is old news. This is in fact the business model taken by some of the leading self-sustaining free software projects like GhostScript, which releases a non-commercial-with-source version and a GPL'd version of GhostScript that's about a year old. Thus, the GPL'd version comes with Linux distributions, non-commercial entities can make use of the latest GhostScript, the company makes money selling licenses to perpetuate the development of the Free Software version, and everyone's happy.

ogg tarkin is somewhat dead.

[Anonymous Coward]

i took a look at how they do things and it doesnt look good.

3d wavelets are useless without motion compensation, they have no motion compensation yet.
even with motion compensation they are gonna have a hard time stitching it together so no artifacts are left over.

tarkin is around since 2000, all they have is some obscure 3d wavelet transform and a huffman backend that leads to good quality at 800kbyte/sec but.. no.. something striped moves by one pixel and everything breaks. yes, its a research codec but comeon.. they have been talking on the mailing list of adapting several other codecs (like vp3) and go on and on over different motion compensation aproaches like meshes and stuff without writing any test code.
so.. as long as no one comes by and drop them a pretty done codec tarkin wont get done.
well ok.. its better then indeo 5 in some cases though.
now that vp3 is lgpld theyll probably just write an ogg header and its done.

Re:ogg tarkin is somewhat dead.

[jonathan_ingram]

Tarkin is not dead.

Tarkin is in the same state that Vorbis was 3 years ago. No-one sensible thinks that it should be competing with MPEG-4/Sorensen/VP* at the moment. No-one connected with the project (only a couple of people, working in their spare time) has been promoting this project as competitive -- only some losers who hype every piece of open source software, no matter how far along in development the software it.

Come back in 2/3 years, and Tarkin will be looking much better.

Re:ogg tarkin is somewhat dead.

[Skuto]

>Question is, will it matter by then? I guess only
>time will tell.

Of course. People will still be encoding video in 3 years. Tarkin technology is angled more towards future development than most formats nowadays which are all more or less based on the level of technology MPEG4 is based on. That doesn't mean it's the last word.

For example, 3 years ago we had MP3. That didn't cause Vorbis 'not to matter'. It is better technology. It is free.

--
GCP

hopes up

[llamalicious]

and here I thought we were talking about virtual pool 3.
don't do that to me so early in the morning, I had sudden visions of making a FFA 8-ball tournament. or 9-ball CTF.

ugh. back to my coffee.

actually it did happen

[paule9984673]

Actually it did happen. Divx networks forked the opendivx code they developed with lots of help from the community, into their proprietary divx4 code.

But some people picked up the opendivx code and kept developping it.

I'd say xvid is about up to par with divx5 now. (Save b-frame support, which is still divx5 only.)

VP3 is overrated.

[inquis]

Doom9's site is the premiere site on the web for video encoding. Doom9 actively tries to get his hands on the newest encoding tools, and periodically he tests them to see which codecs give the best results.

It used to be that along with the lastest versions of DivX, he tested WMV and VP3; he doesn't anymore: WMV and VP3 consistently lost and lost badly to div3 sbc, div4, and xvid. You can't say that VP3 is "the next DivX" when it's can't outperform the ancient div3, much less div4 or the newly released div5 / xvid.

[shameless plug]

I really, really, /really/ like xvid. It's an open source reimplementation of Project Mayo, the project that led to the development of div4. Development is fast; I have realized significant gains in quality and usability in even the past two weeks. The codec is fast; on my crappy windows machine it crunches frames faster than div4 and div5 and its playback filter (w/postprocessing!) uses fewer CPU cycles than div4 or div5's.

If you want to play around with xvid, the easiest way to start is to go to the xvid forums at doom9.org and read about what the codec can do for you.

[/shameless plug]

-inq

Next generation is NOT under LGPL/GPL.

[Svartalf]

They're selling the VP4 codec and have a VP5 that's in testing right now. They plan on selling that- if they get an open sourced version of the prior generation out there that can be supported by their newer codecs, that's a win for most people (So long as they provide the decoder for the latest format for free, that is... :-) I'll be posing that question to them on the list. It's great and all, but unless we've got decoders for VP4 and VP5, it's not as good as it could be.

What license was opendivx under?

[Svartalf]

I can't remember...

If it was the LGPL, Divx Networks may be in trouble as they have to honor the license grant given to them for the submitted code (LGPL)- it means that they could be sued bu the submitters for Copyright infringement. If it was under a BSD-ish or X11 license, they would be perfectly fine to do what they did. (Legally, that is- right and wrong doesn't even enter into this at this point...)

Not so...

[Svartalf]

Yes, it may not perform as well as xvid, but xvid has a severe drawback, that you should be giving consideration to- patent licenses.

There is not an aspect of MPEG4 (which is what xvid is a codec for...) that is not covered by patents that need to be licensed. Nobody has a royalty free license to any MPEG4 patents for free software implementations so distribution of the codec is violating the patent rights of several companies.

You may not care now, but they're stepping up enforcement efforts of all this stuff and you may well find yourself without a codec or in trouble because you're violating the rights too.

Re:BSD is better how? (Ogg uses BSD)

[sab39]

I'm citing RMS's stated opinion on this, but it happens to be one I agree with (actually I don't usually really care about what license anything's under so long as it's truly Free/OpenSource according to the respective definitions).

In the case of Ogg, this is a file format we are talking about. The value of a file format isn't proportional to the quality of the format, but to the amount of content available in it. The amount of content available will be determined by how many people use it.

Freedom to view content in a file format that nobody uses is not much use, so in this kind of situation insisting on preserving freedom to the detriment of wide usage doesn't gain anything. Therefore, in this situation it's better to take the hit of getting our (that is, the free software community's) code used in evil proprietary software than to insist on complete freedom and have many corps not use the format.

RMS is dogmatic about many things, but he is willing to be pragmatic in certain cases. Generally his rule seems to be that he'll make pragmatic decisions if they (1) don't compromize his ethics, and (2) result in a *net gain* of freedom for software users.

Stuart.

XviD as alternative, Ogg Tarkin in the future

[tangent3]

The problem with Ogg Tarkin it is still pretty much an experiment, using techniques which is way ahead of its time. 3d wavelets haven't been heard of in any other standard which are under development.

XviD, however, exists today. It is a fully GPL'ed MPEG4 codec. However, it cannot exist legally in any form other than an experiment because the MPEG4 license still has to be paid for in order to use XviD. XviD, like LAME, will mostly exist as CVS sourcecodes under guise as an experiment, with many rogue sites around the world providing binaries (usually with automated daily compiling).

Personally, I wouldn't count on the quality of VP4 being anything earthshattering. Tests of VP3 quite clearly shows that it is behind DivX3, DivX4 and XviD in terms of quality, so something has to really shape up. This might probably be due to a lack of 2-pass VBR encoding feature in VP3. Meanwhile I will just continue to encode my rips in XviD, encode the audio with Vorbis and mux them together into an Ogg container. If VP4 or VP5 really turns out to be good enough, I would probably try to find a way to mux that video stream into the Ogg container too.

Re:XviD as alternative, Ogg Tarkin in the future

[JoeBuck]

The only way that Tarkin can hope to be patent-free is either by doing something completely different or by using approaches that were known 20 years ago; all the standard approaches are covered by patents.

Re:XviD as alternative, Ogg Tarkin in the future

[Pathwalker]

I wouldn't have much faith in the tests you link to. I looked at the settings that the tester used, and it looks to me like they deliberately crippled vp3.

To quote the article:

"In VP3 I set the bitrate, keyframe interval to 9999 seconds, auto keyframe turned on as well as quick compression"

In my experience VP3 only gets noticibly blocky (the tester's major complaint) when it is prevented from creating a keyframe when it wants to. Here, they pretty much prevented vp3 from generating keyframes at all. The keyframe interval should have been left BLANK not set to a stupidly high number.

Additionaly, there is another menu of keyframing options (the one he should have used to set the adaptive keyframe rate rather than locking it) of which he writes nothing. Here, I probably would have set the minimum time to about 1/4-1/2 second, and set the maximum time to the highest supported number.

Furthermore, There is an image quality control which controls the tradeoff between image quality, and the risk of dropping the frame rate. No mention was made of the setting of this control, but the complaints about low detail make me wonder what it was set to.

Finally, turning quick compress on does lower quality. For a test which did not involve encoding speed, I have to wonder why the tester chose to turn that option on, as it trades off quality for faster encoding!


I use vp3 to encode DV streams (in Quicktime) for viewing over the web. Vp3 is a very good quality codec, superior in many cases (unless you are streaming from a QTSS, or the source was shot under unusual light conditions) to the free version of Sorenson. It is excellent under these conditions.

Not really streaming, but progressive download

[benwaggoner]

A nitpicky point - VP3 as implemented in AVI and QuickTime files is designed for progressive download, not true, real-time streaming. Thus, you get the classic movie trailer wait-awhile-and-play experience, but without the ability to do random access over long files and that kind of groovy stuff.

Good support for real-time streaming would require a native packetizer to build a hint track that the (open source) Darwin Streaming Server uses to determine packetization of the stream, and which helps loss recovery and other good stuff.

Adding a native packetizer for VP3 would be an excellent open source project for the codec.

Re:Another nitpick

[Wesley Felter]

Wrong; the APSL is listed as an open source license at opensource.org. Or were you talking about some other kind of open source?

Huh?

[SpinyNorman]

The doom9 site only tested VP3, not VP4 nor the latest VP5 -- they're 2 whole CODEC generations behind.....