• 1
Will Xiph collaborate? The VP8 encoder is in dire need of activity masking and temporal RDO :)
(Frozen) (Thread)

Yes, we'll be collaborating. Already have been, in fact, for the past few weeks.
(Frozen) (Parent) (Thread)

What can you said Monty about the comments of an x264 Developer: http://x264dev.multimedia.cx/?p=377

Do you believe that VP8 is more or less a rip off h264, that there must be patents and that the specs "is imprecise, unclear, and overly short, leaving many portions of the format very vaguely explained."

What are your comments about choosing Matroska instead of Ogg? Isn't WebM just a subset of Matroska anyway? Why Ogg don't fit in Google plans?
(Frozen) (Thread)

Re: What can you said about...

"Do you believe that VP8 is more or less a rip off h264"

No. That's pretty serious hyperbole. Some design chunks look very similar even to the trained eye, but details matter in both code and patents. DS ignored the details.

"that there must be patents"

If any litigation materializes, I doubt it will be motivated by actual infringement. Here's where it's nice to have Google around.

"is imprecise, unclear, and overly short, leaving many portions of the format very vaguely explained."

Sadly, he's correct about this. We can fix the spec and write a good one after the fact, but this leaves up in the air how many bugs are undiscovered because no one documented the hard parts (so no one else could check them). Tim's been pouring over the code for weeks already and has caught a few such instances, I don't think any were major.

That said, despite Google's claim that things are already frozen, I'm sure that if a showstopper popped up they'd change their minds. I don't think there are any showstoppers. What we'll probably find in the future are instances of 'aw, geez, it would have been nice to change that if it wasn't too late.' We have some amount of experience dealing with that from Theora :-)

Edited at 2010-05-20 07:19 am (UTC)
(Frozen) (Parent) (Thread)

Re: What can you said about...

(Anonymous)
And about the quality of the codec itself? Any preliminary comment about where it stands now and where it can get? (Is it competitive/better with good h264 implementations? Will it be? Is it really a better alternative than Theora?)
(Frozen) (Parent) (Thread)

Re: What can you said about...

Quality of the format: Theora was good enough and VP8 is solidly better than Theora (even current Ptalar). So, awesome, it's a pretty pure win.

I know everyone wants a 'but how does it compare to h.264' answer, or more specifically 'how does it compare to x264'. A: Total red herring; none of the orgs with money involved care much about that answer. It's only good for trash talking, not actually winning a fight. B: I don't actually know for sure yet. We've been concentrating on evaluating the spec and the codebase quality rather than drag racing it, because pinning VP8's success on a size contest completely misses the point.

Theora was already beating some h.264 encoders out there (no, not x.264) and VP8 at very least already beats more. As far as the sourcebase goes, it raises the ceiling over Theora on how far we can improve things. Neither h.264 nor VP8 will be holding still.

...but once I have a clearer picture of the answer to B, I'll still fess up ;-) Chances are the answer is 'about the same', at least right now. We'll see.
(Frozen) (Parent) (Thread)

Re: What can you said about...

"What are your comments about choosing Matroska instead of Ogg?"

Don't really have any--- they were equally good for this particular task and they chose Matroska not Ogg.

"Isn't WebM just a subset of Matroska anyway?"

Yes. I think the EBML identifier might be different.

"Why Ogg don't fit in Google plans?"

I can speculate, but I wouldn't necessarily be right. I think the decision was relatively arbitrary.
(Frozen) (Parent) (Thread)

Re: What can you said about...

(Anonymous)
Are the Ptalarbvorm developers now working on VP8, instead?
(Frozen) (Parent) (Thread)

Re: What can you said about...

In addition to, not instead. Theora is here now, VP8 is still arriving. Knowledge gained is likely to be transferrable.
(Frozen) (Parent) (Thread)

Hooray! This is excellent news!
(Frozen) (Thread)

Do the existing H.264 hardware acceleration primitives implemented on devices like phones work to speed VP8, or will that require new silicon?
(Frozen) (Thread)

The vast majority of 'hardware' h.264 implementations are just embedded general purpose DSPs. Theora had already gone onto a few, VP8 can be added the same way.
(Frozen) (Parent) (Thread)

  • 1
?

Log in