Re:The first in-depth technical analysis of VP8
by naner on Wednesday May 19, 2010 @01:32PM
(#1361654)
> That said, I've been reading through the libvpx source for the past 20 minutes or so. Jason repeatedly slams the quality of the source. The code is much cleaner than he suggests and rather well-commented.
It is possible he had an older revision of their code. Since he didn't say how he obtained it early, it may have not been through official channels.
Re:The first in-depth technical analysis of VP8
by DarkShikari on Wednesday May 19, 2010 @01:53PM
(#1361739)
I had rc8 through an NDA which has now expired (hence why I waited until now to post the analysis).
Re:The first in-depth technical analysis of VP8
by mpk on Wednesday May 19, 2010 @09:17PM
(#1362946)
> It is possible he had an older revision of their code.
Unless he had a very ancient version of this codebase it is unlikely that it is different in structure from the current mainline.Adding comments or sanitizing local variable names are only very superficial changes.
Re:The first in-depth technical analysis of VP8
by rpdillon on Thursday May 20, 2010 @04:03AM
(#1363494)
Except that he goes out of his way to slam the comment quality. Besides, I think the difference between well-commented code and code without comments can be quite vast, especially when you're inlining assembly, as is the case here.
Re:The first in-depth technical analysis of VP8
by kevingadd on Wednesday May 19, 2010 @01:27PM
(#1361639)
I think it's reasonable to be overly critical of the source code quality based on the factors he cited (the lack of a real spec and the extreme age of the codebase) - if the code is the only thing we have to go by in many cases, it had better be pristine.
Hopefully there's some sort of a real specification forthcoming and it just isn't ready for release yet.
While I only have a surface-level understanding of video encoding and compression techniques, there are some very good criticisms here. That said, I've been reading through the libvpx source for the past 20 minutes or so. Jason repeatedly slams the quality of the source. The code is much cleaner than he suggests and rather well-commented.
Better than H.264/main or not, I'm excited to see where VP8 goes. With implementations in Chrome, Firefox, Opera, and Flash, it's off to a good start from day one. We'll see where it heads from there.