libvpx 1.4.0 ("Indian Runner Duck") officially released

[xiphmont]

Johann Koenig just posted that the long awaited libvpx 1.4.0 is officially released.

"Much like the Indian Runner Duck, the theme of this release is less bugs, less bloat and more speed...."

Comments

[(Anonymous)]

Do the browsers (plan to) support 10/12-bit color? What about 4:2:2 & 4:4:4 chroma? I know they did for Theora. Also I know it's taboo on browser land, but has any thought been given to dropping earlier codec support (VP8, Theora, maybe even Vorbis)?

[(Anonymous)]

I believe that browsers just integrate libvpx and so on, so they'll get whatever the underlying libraries support.

I don't think that it is a good idea to deprecate any open video codecs until the battle against H265 has been fully won. Then audio/video codecs should be treated like any other HTML feature in terms of support/deprecation. We need to keep the pressure on Apple an Microsoft to integrate VP9 into their browsers until then...

[(Anonymous)]

We need to keep the pressure on Apple an Microsoft to integrate VP9 into their browsers

I hope they integrate VP9, but they probably won't. I think Apple and Microsoft see VP9 (and VP10 if it's released in future) as Google's codec which is almost exclusively built for use on YouTube to reduce bandwidth and avoid streaming royalties. They may be right.

When Google talks about releasing a new, and presumably bitstream incompatible, VPX codec every 18 months (http://www.cnet.com/news/googles-web-video-ambitions-run-into-industry-reality/) it seems clear that they have no ambition to build an eco-system around the VPX line of codecs. At that rate of codec churn, third parties are not going to be much interested in getting involved with VPX.

I think the best pressure to apply to Apple and Microsoft would be to ask them to get involved with and contribute to the NetVC effort. A standardized, high quality, royalty-free video codec with long term support is what the Web needs. Opus delivers on that for audio and maybe NetVC can deliver it for video.

[(Anonymous)]

It's not that obvious, they have to hook the corresponding colorspaces and such.

If this file's name is to be believed http://downloads.webmproject.org/test_data/libvpx/vp93-2-20-12bit-yuv444.webm then yes they do, but Firefox shows a green artifact on it (maybe bug 549767 or something similar), and I suspect Chrome is downconverting to 4:2:0 (it's difficult to tell on that sample though).

[(Anonymous)]

It's not that obvious, they have to hook the corresponding colorspaces and such.

If this file's name is to be believed http://downloads.webmproject.org/test_data/libvpx/vp93-2-20-12bit-yuv444.webm then yes they do, but Firefox shows a green artifact on it (maybe bug 549767 or something similar), and I suspect Chrome is downconverting to 4:2:0 (it's difficult to tell on that sample though).

[(Anonymous)]

It's not that obvious, they have to hook the corresponding colorspaces and such.

If this file's name is to be believed downloads.webmproject.org/test_data/libvpx/vp93-2-20-12bit-yuv444.webm then yes they do, but Firefox shows a green artifact on it (maybe bug 549767 or something similar), and I suspect Chrome is downconverting to 4:2:0 (it's difficult to tell on that sample though).

[(Anonymous)]

It's not that obvious, they have to hook the corresponding colorspaces and such.

If this (1) file's name is to be believed then yes they do, but Firefox shows a green artifact on it (maybe bug 549767 or something similar), and I suspect Chrome is downconverting to 4:2:0 (it's difficult to tell on that sample though).

(1) downloads webmproject org/test_data/libvpx/vp93-2-20-12bit-yuv444.webm

[(Anonymous)]

(actually bug 562858)