Why not WebP use VP9?

[doraz...@gmail.com]

As the testing from xiph.org, VP9's still image codec is more better than VP8's.:

https://people.xiph.org/~xiphmont/demo/daala/update1.shtml

[Pascal Massimino]

Hi,

On Wed, Jun 21, 2017 at 11:12 PM, <doraz...@gmail.com> wrote:

As the testing from xiph.org, VP9's still image codec is more better than VP8's.:

https://people.xiph.org/~xiphmont/demo/daala/update1.shtml

There's no simple answer to this. Some points to make:

  - on the above page, 'VP8' probably means: the *video* codec. 'libwebp' is re-implementation of the encoder targeting still image coding, which is a different problem than video coding. The encoder doesn't make the same decisions, or focus on the same features.

  - when WebP was launched, it was agreed that the codec for lossy compression (VP8 at that time) would not change afterward. Even if some better codec would certainly emerge. This was to ensure stability (you can't change the format every years or so, each time something new comes out).

  - WebP sticked to VP8 in order to be able, one day, to use hardware decoders for lossy decoding.

Going forward, a new -more efficient- image codec aiming at the same target than WebP (online image traffic) would rather use the latest free codec (AV1) instead of VP9. Using some parts of AV1 (along with other new ones for lossless, alpha, etc.) for a new image codec is an idea floating in the air lately. That's definitely of interest...

 

hope it helps,

skal/