Multimedia 2006 : One World, One Standard.

[Musafir_86]

Greetings everybody,

-As you can see my topic title is copied from Beijing 2008 : "One World One Dream" slogan. ;)

-Okay, the truth is I'm suggesting that we all should consider using one standard in digital multimedia contents (digital audio/video). I know we're humans, & human have varieties, but it is wrong for us to use one format for the sake of our generation (& us too) in the future? :D

-So, I'm calling OSS developers from around the world (who currently in charge of XviD, x264, ffmpeg, LAME, FAAC, FAAD2, CoreAAC, etc.) to contribute your knowledges, skills, eXPeriences, (& all you have) to these :

->For audio, Ogg Vorbis;
->for video, Theora/Snow;
->and for container, Matroska.

-The reason I choose all those formats is because they're ROYALTY & PATENT FREE. So, it would save us (the consumer) and the (commercial) developers/hardware manufacturers too. Also, it would avoid problems like BetaMAX vs VHS, DVD+R/RW vs DVD-R/RW vs DVD-RAM, HD-DVD vs Blu-Ray, etc; which make us (consumers) suffer in the end. :(

-After all, this is my opinion (and hope). It's up to you all to accept it (or not). Anyway, keep those good works going on! :)

Thanks again & bye.

Regards,
-Musafir_86.

[acidsex]

Dude, no way that will ever happen, thank god. Choice is what life is all about. Besides, neither of your proposed codecs come close to whats available quality wise today.

There is no way Theora will ever come close to what Ateme has developed thus far with their H.264 codec and its still maturing quickly. No way the Vorbis group will catch them since most of the developers have other jobs for income since no funds are derived from their projects.

Nothing wrong with having lots of choices. The quicker you accept it, the better off you will be. Each codec serves its own purpose for different users. If you like your codec, use it. But dont ever expect serious playback on set top boxes any time soon.

P.S. I am not dumping on you just stating me opinion and resistance to your "standard" request.

[Musafir_86]

-That's okay, but how about Dirac (http://dirac.sourceforge.net)? :p

[superdump]

Dude, no way that will ever happen, thank god. Choice is what life is all about.
Agreed.

Besides, neither of your proposed codecs come close to whats available quality wise today.
Not agreed. Snow is exceptionally good. Progress has slowed over the last month but it's still right up there. It's a little soft for my liking though and there are slightly noticable chroma problems which no-one seems to know how to fix, but still, it's very very good. Theora is a bag of crap imo currently however.

There is no way Theora will ever come close to what Ateme has developed thus far with their H.264 codec and its still maturing quickly. No way the Vorbis group will catch them since most of the developers have other jobs for income since no funds are derived from their projects.
Vorbis is very good anyway and aoTuV does some tuning work on it from time to time and I think his/her tunings are very good, I could back this up with blind test results from the HydrogenAudio community. There's no doubting how good Ateme/Nero's codecs are though.

EDIT: From what I've read and heard Dirac is much slower than Snow and not as good quality-wise anyway.

[Joe Fenton]

Who says snow is patent free? It uses wavelets from what I remember from the threads. ANYTHING using wavelets WILL violate several hundred patents. The last time I did a check on the USPTO archive, there were nearly four thousand granted patents on using wavelets to process video.

[morsa]

using wavelets doesn't strictly mean that you voilate any patent.
Many transforms usefull for wavelets are so old as to be imposible to have any patent on them.
It's like sayin DCT violates a patent.Yes it is patented but its patent is rather old, so it doesn't apply anymore...
Same goes for Hadamard matrix.huffman, Arithmetic coding and many other several things...
"Software patents serve one purpose:Slow further development for at least 25 years"

[Leak]

Who says snow is patent free? It uses wavelets from what I remember from the threads. ANYTHING using wavelets WILL violate several hundred patents.

Well, Dirac is using wavelets as well, but was designed not to clash with any patents, so I rather doubt that last sentence of yours...

np: Plaid - Dial P (Dial P)

[bond]

one such standard would be indeed great and some might remember that i once proposed the "bond standard" ;)

using vorbis as audio is definitely a good idea cause its free and offers great quality
for video dirac and theora are also good, theora definitely has the potential to play in the league of the good codecs, the only problem is there is no good implementation of theora. dunno the status of dirac
about matroska, i am not sure if matroska is "patent free", but its definitely a great container
what sub format should be used?

actually i have to say that xiph already offers a lot of important things already: vorbis, theora, the ogg container and writ as the subtitle format
its just that hardly anyone really realises this and starts to work with the formats

actually i think that ratDVD would have had the chance to become such a "standard", but the big problem is that their video codec is totally crap and i still wonder why they chose to develop an own one...

[Tommy Carrot]

Come on now, in the era of h.264 who would go back to theora? I know you have different opinion, but imo theora is clearly inferior to ASP theoretically too (no qpel, no b-frames, etc.), not just implementationally. Imo very few people would actually want to use it.

Dirac is more promising, theoretically it's very powerful, the problem lies in the current implementation: bad quality, strange artifacts, slow speed, etc. But at least it has some hope to slowly evolve into something competitive.

[bond]

Come on now, in the era of h.264 who would go back to theora? I know you have different opinion, but imo theora is clearly inferior to ASP theoretically too (no qpel, no b-frames, etc.), not just implementationally. Imo very few people would actually want to use it.
did you read the specs? theora offers lots of things which make it better than asp: loop, multiple reference frames, 8x8 blocksizes, more cqms, better entropy coding
qpel and b-frames is the only big thing it lacks compared to asp, and on2 demonstrates nicely that b-frames are not a must at all

still i dont say that its better than avc, cause it isnt

also an important point of this discussion was to use free formats and avc is surely not one of them

[Tommy Carrot]

I don't regard loop-filter as an absolute improvement, it blurs the image pretty badly, and most users are disabling the post-filtering for ASP anyway. Imo multiple reference frames (more precisely 2 ref frames) doesn't have too big impact on quality in the case of h.264, i don't think it will have with theora. I have to admit i don't really get why 8x8 blocksizes is an advantage, as ASP uses it too.

Now the ability to use more cqm and adaptive huffman coding are indeed improvements, but i don't think they can offset the lack of b-frame and qpel options.

And yes, vp7 (and snow) are quite good without bframes too, but they could be even better with it. :)

EDIT: i know AVC is not free, but this is not factor for most people. And i believe Dirac will be more adequate as a free codec (well, if the development continues at this pace, we will have to wait for a long while for a good implementation... but this is true for theora as well).

[acidsex]

-The reason I choose all those formats is because they're ROYALTY & PATENT FREE. So, it would save us (the consumer) and the (commercial) developers/hardware manufacturers too. Also, it would avoid problems like BetaMAX vs VHS, DVD+R/RW vs DVD-R/RW vs DVD-RAM, HD-DVD vs Blu-Ray, etc; which make us (consumers) suffer in the end.


The reason I gave my answer was because of this part of the original post. Its never going to happen mainly because NONE of us are in any position to dictate how content will be distributed. One has to figure that Hollywood will determine whats best for them (cost, DRM, etc...) and you better believe that they could give a crap what OSS developers or those of us that want to make backup copies want.

Again, each codec offers something for most people that the other codecs dont. My example is one I posted on another thread. I was able to encode a 15 hour video tutorial 880x660 at 5fps at 2.4GB and there was no visible quality loss with WM9. I was unable to do the same H,264 AVC as the size doubled and the quality was almost as good as the WM9 output.

The consumers which he references above, will not decide what wins. Hollywood does and will. Consumers will have two choices. Either buy the content on the media (HD-DVD/BluRay/DVD/etc...) and the codecs they choose (Mpeg-4, WM9, MPeg-2, H.264 AVC/etc.) or dont buy it and not own the content you want.

The consumers will have some impact but they will not force studios into what they want. While many here are well versed in video/audio encoding and the various choices in codecs, your average consumer is compeltely clueless. The majority just want to stick a disc in their player and have it play. They care nothing about how sexy H.264 AVC is or that M$ isnt open source. Cost, content, and ease of use is what will drive the market.

Again, while it would be nice if everyone settled on one video standard, one audio standard, one media standard, etc... there would be no reason for developers to try and improve the product. One needs to look no further than M$ and its operating system for an example. How innovative are they really? Let smaller companies develop successful programs and then M$ implements them in their OS and call it revolutionary. Please!

Technology will constantly change and innovations will get better. It was just a few years ago where audio lovers considered mp3 the holy grail. Now mp3 gets its ass kicked by AAC in almost every test I have seen.


which make us (consumers) suffer in the end.

Sadly, consumers arent what studios care about. We are not green enough and cant fit into their fat wallets. Maybe someday that will change but dont hold your breath. If the studios cared about consumers, they wouldnt charge consumers for a "digitally rematered" cd from the late 80s that they were overcharge for then. So to get what the consumer should have received in their first place, they had to pay twice. Or how about someone who buys a DVD today and then in three years, they release a directors cut or a special version and a fan would have buy another one. Its all about money and money is exactly why you will never see one standard in our life time if ever.

[On2Tech]

[QUOTE=Tommy Carrot]

And yes, vp7 (and snow) are quite good without bframes too, but they could be even better with it. :)

QUOTE]

The issue with B-Frames is that there are a LOT of patents in this area, which is why we steered clear of them in Vp3, 4, 5, 6 and 7 and why they are not present in Theora (which is of course based on VP3).

Until these expire, this probably rules out their use for all royalty free or open source codecs.

[Tommy Carrot]

The issue with B-Frames is that there are a LOT of patents in this area, which is why we steered clear of them in Vp3, 4, 5, 6 and 7 and why they are not present in Theora (which is of course based on VP3).
Hmm, i wasn't considering that, because even mpeg 1 uses b-frames, and it's quite an old compression standard, i thought the related patents are already expired.
Until these expire, this probably rules out their use for all royalty free or open source codecs.
Dirac has a feature (L2 frame), which is something like an improved, more flexible b-frame, and dirac is supposed to be royalty-free.

[On2Tech]

Hmm, i wasn't considering that, because even mpeg 1 uses b-frames, and it's quite an old compression standard, i thought the related patents are already expired.

Dirac has a feature (L2 frame), which is something like an improved, more flexible b-frame, and dirac is supposed to be royalty-free.

Many of the B frame patents do date from late 1980's and early 1990s (around the time Mpeg-1 came out). Not sure of the expiry dates on all of them but some are certainly still in force and included in the MPEG-LA portfolios.

We talked with an industry expert on video patents a while back and he advised then that anything that used Bi-directional prediction in any meaningfull way was likely to infringe on one or more of these patents.

I cannot speak for what Dirac does but it was something that we felt we could not risk at the time we were designing the feature set for VP7.

[Doom9]

So, it would save us (the consumer) and the (commercial) developers/hardware manufacturers too.You're certainly not considering the companies that make money licensing patented technology (whether or not we'de be much better off without a lot of these patents in general is another thing entirely). So it's really a point of view.

For a private user, in the video arena licensing fees do not really matter. You can get excellent open source codecs, otherwise free codecs, or some for a modest fee (like DivX, VP7). And if you are into business, you should pick the best codec for your customer.. which is not necessarly the best for your wallet. If you're streaming audio, certainly Vorbis is an option. But in video, the codecs you listed just cannot compete. Theora looks like crap, Snow is slow and has problems, Dirac is beyond usable so most people can't even comment on quality because they never get to the point where they have a playable video.

And we've had the "developers should do so and so" discussion. There's no need to bring that up again.

Everybody should pick what works best for them, users and developers.

[mimungr]

I think the vast number of audio/video/container patents makes it impossible for anyone to claim with confidence that their technology doesn't infringe. Look at the lengthy process ISO went through trying to track down all the patent holders for MPEG-4. If any of the "patent free" codecs started to generate serious revenue, you can bet someone would slap them with a lawsuit. Such claims are a daily occurrence in the software industry.

[bond]

claims dont necessarily have to be true