Google Opens Up VP8, Launches New Container Format

[smitty3268]

Actually you are wrong there. As a entity it can sue. They did just that agains Alcatel.

http://www.mpegla.com/main/pages/media.aspx

Hmm, that's very interesting. When companies sign up are they forced to give over the right to sue like that? Or did the MPEG-LA just sue at the behest of the companies that wanted them to?

IP law always gives me a headache...

[deanjo]

That would be because you're converting a h264 (lossy) into a V8 (lossy) = more lossy?

Correctly comparing codecs requires, at the very least, compressing both from the same source, and to be more fair the source needs to be very raw.

Definitely interested in some comparison tests.

I've done just that already btw. Take from a pure source (BBB non compressed frames in this case) and encoded vp8 at it's highest quality and then compared them to a h264 encode at the same file size (didn't even tweak the settings in HB) the results were that h264 still has a very visible advantage in terms of quality.

[deanjo]

Hmm, that's very interesting. When companies sign up are they forced to give over the right to sue like that? Or did the MPEG-LA just sue at the behest of the companies that wanted them to?

IP law always gives me a headache...

No actually the situation is that the MPEG-LA will sue on behalf of their members. When a company puts a patent into the MPEG-LA pool it becomes more or less the owner of that IP. Settlements are then put back into the consortium. Much like when the Free software foundation sues.

[smitty3268]

I've done just that already btw. Take from a pure source (BBB non compressed frames in this case) and encoded vp8 at it's highest quality and then compared them to a h264 encode at the same file size (didn't even tweak the settings in HB) the results were that h264 still has a very visible advantage in terms of quality.

h264 baseline, or were you using higher settings? I don't think YouTube even uses b-frames in their h264 encodes, so that's one big advantage h264 has that doesn't even matter in that case.

Not that i'd argue the point that x264 would be superior. The argument is that VP8 is "good enough", not better.

[deanjo]

h264 baseline, or were you using higher settings? I don't think YouTube even uses b-frames in their h264 encodes, so that's one big advantage h264 has that doesn't even matter in that case.

Not that i'd argue the point that x264 would be superior. The argument is that VP8 is "good enough", not better.

Encoded with baseline. Even badaboom encodes provide a better result.

[smitty3268]

Even badaboom encodes provide a better result.

Ouch. Too blurry? Hopefully they can just remove some of those PSNR optimizations, which i think are partly to blame.

[deanjo]

Ouch. Too blurry? Hopefully they can just remove some of those PSNR optimizations, which i think are partly to blame.

Blurry and macroblocking, hopefully those can be resolved.

[deanjo]

Just a personal experience to share. We send out training videos to our clients on a daily basis. The prospect of having a open codec to render our training videos in does appeal to us. These videos are done with Camtasia by the trainers who record their training sessions. Trying the various codecs over the last day I can say with confidence that VP8 will need a lot more work to meet our needs. Right now we use windows media 9 for the simple reason that all of our clients have it already on their system. For that use it does a very good job. Encoding some of the video's however in VP8 even at highest quality makes the text on the screen captures very blurry. These are not complex items to render into video. If anything they are the perfect scenario for video codecs that should present high compressibility with minimal detail loss.

[monraaf]

That would be because you're converting a h264 (lossy) into a V8 (lossy) = more lossy?

Correctly comparing codecs requires, at the very least, compressing both from the same source, and to be more fair the source needs to be very raw.

Definitely interested in some comparison tests.

I think comparisons have already indicated that h264 encoded with x264 can have better image quality than VP8. My point is merely that for the typical user watching YouTube videos it doesn't matter quality wise if they are watching the h264 version in a flash player or the VP8 version native. I added two screen shots, nothing scientific but just to give you an idea. And of course I encourage anyone to try it out the webm videos themselves.

[deanjo]

I think comparisons have already indicated that h264 encoded with x264 can have better image quality than VP8. My point is merely that for the typical user watching YouTube videos it doesn't matter quality wise if they are watching the h264 version in a flash player or the VP8 version native. I added two screen shots, nothing scientific but just to give you an idea. And of course I encourage anyone to try it out the webm videos themselves.

The amazing thing is that I can tell the difference from just the thumbnails alone. Notice the over contrast on the pic on the right?