Linked by Thom Holwerda on Wed 14th Mar 2012 19:37 UTC
Internet & Networking Ever since it became clear that Google was not going to push WebM as hard as they should have, the day would come that Mozilla would be forced to abandon its ideals because the large technology companies don't care about an open, unencumbered web. No decision has been made just yet, but Mozilla is taking its first strides to adding support for the native H.264 codecs installed on users' mobile systems. See it as a thank you to Mozilla for all they've done for the web.
Thread beginning with comment 510677
To view parent comment, click here.
To read all comments associated with this story, please click here.
RE[3]: In through the back door.
by lemur2 on Thu 15th Mar 2012 08:33 UTC in reply to "RE[2]: In through the back door."
lemur2
Member since:
2007-02-17

Initially there were some extensive tests on comparing VP8 and H264, have they been re-done recently with the results posted online for us to check? Do you have any proof that VP8 is better than H264?

Otherwise, even Google is not claiming that VP8 is better than H264, just that they've improved by a certain percentage on a certain metric.


One can unfortunately debate this topic ad infinitum. About a year ago, after WebM quality had twice been significantly improved compared to its initial release, on OSNews there was a topic about encoder speed, and a poster who knew how to do so provided a couple of examples using WebM and h264 encoder profiles such that the quality of the output was all but identical. As close as could be achieved.

The x264 encoder was far faster. The WebM files were actually a bit smaller.

Google have significantly improved the encoder speed of WebM twice since then.

http://blog.webmproject.org/2011/08/vp8-codec-sdk-cayuga-released.h...

http://blog.webmproject.org/2012/01/vp8-codec-sdk-duclair-released....

I think that the x264 encoder speed is still faster than libvpx, but nevertheless the point remains that the WebM format actually needs slightly fewer bits for the same quality of output.

Edited 2012-03-15 08:36 UTC

Reply Parent Score: 2

saynte Member since:
2007-12-10

You can't convince anyone that VP8 gives more quality per bit than H264 in that way. A convincing comparison would give both quantitative and qualitative points, like some of the early ones comparing PSNR and SSIM, along with screenshots of the same frame encoded.

The last known tests like this (compression.ru, as well as other sites) put H264 ahead of VP8 in most every category. You'd have to retest with the updated encoders to be sure.

Reply Parent Score: 2

lemur2 Member since:
2007-02-17

You can't convince anyone that VP8 gives more quality per bit than H264 in that way. A convincing comparison would give both quantitative and qualitative points, like some of the early ones comparing PSNR and SSIM, along with screenshots of the same frame encoded.


Why not? You eliminate one of the variables (the quality, by selecting profiles which yield the same quality, no matter what happens to be the name of the profiles), and then you can compare the other variables, being file size and encoding speed. It turns out that WebM has slightly better quality per bit, and x264 is significantly faster, for encoding to the same quality.

The last known tests like this (compression.ru, as well as other sites) put H264 ahead of VP8 in most every category. You'd have to retest with the updated encoders to be sure.


They were done ages ago. The WebM codec has been improved very significantly in both speed and quality-per-bit in each of four releases since then.

Reply Parent Score: 2