Tweet
It's impossible to be certain about what the market will do.
I'm not sure about this. For one thing, a simple addition of B-frames would increase the efficiency of VP8 by about 20% (according to that x264 dev). B-frames might be patented, but I would be surprised since it is such an obvious extension to P-frames. Now, while that would certainly break the binary, so what? How many hardware decoders do they have? Speaking of hardware, did you see the list of partners Google assembled? http://www.webmproject.org/about/supporters/
That's a who's who of the mobile hardware sector. Marvell/Qualcomm/Imagination, there you have the guys who can create the dsps needed for the devices. It's a matter of time, really, and there's not a huge hurry, I don't think. But, you might mention the next gen codec successor to m4p10? Ars had an interesting article about codecs a few months back. It was thought that we are rapidly approaching diminishing returns (which one could say we've reached already with h264 which, while better than ASP, is much harder to decode) and pretty much the best we could expect from the next codec would be a 20% improvement, and that comes at a tremendous cost in complexity. I don't know how true that is, but I don't recall hearing anyone arguing that number from the comments. If you know more I'd be interested in hearing about it since there's not a ton of info out there.
As for MP3, that was created back in the early 90's I believe. Vorbis didn't have a stable bitstream until 2000 and software didn't exist until 2002. This was too late, plus they had nobody behind them, to my knowledge.
Also, VP8, by it's definition, should be easier to decode than h264 so it may not be impossible to simply play the file directly on the processor (probably need to actually make use of the NEON extensions, though). This should work until Goog's partners have had enough time to shipout the new DSPs .
I'm certainly not saying that VP8 is a sure thing, but it's got as good a backing as one can hope, and from the reference pics that were shown on the x264 blog, the difference wasn't tremedous, IMHO.
Best/Liam
Originally posted by deanjo
View Post
That's a who's who of the mobile hardware sector. Marvell/Qualcomm/Imagination, there you have the guys who can create the dsps needed for the devices. It's a matter of time, really, and there's not a huge hurry, I don't think. But, you might mention the next gen codec successor to m4p10? Ars had an interesting article about codecs a few months back. It was thought that we are rapidly approaching diminishing returns (which one could say we've reached already with h264 which, while better than ASP, is much harder to decode) and pretty much the best we could expect from the next codec would be a 20% improvement, and that comes at a tremendous cost in complexity. I don't know how true that is, but I don't recall hearing anyone arguing that number from the comments. If you know more I'd be interested in hearing about it since there's not a ton of info out there.
As for MP3, that was created back in the early 90's I believe. Vorbis didn't have a stable bitstream until 2000 and software didn't exist until 2002. This was too late, plus they had nobody behind them, to my knowledge.
Also, VP8, by it's definition, should be easier to decode than h264 so it may not be impossible to simply play the file directly on the processor (probably need to actually make use of the NEON extensions, though). This should work until Goog's partners have had enough time to shipout the new DSPs .
I'm certainly not saying that VP8 is a sure thing, but it's got as good a backing as one can hope, and from the reference pics that were shown on the x264 blog, the difference wasn't tremedous, IMHO.
Best/Liam
Comment