Google VP9 "Next Generation Open Video" information posted

[Keiyakusha]

Originally Posted by Rumbah

Photoshop saves jpegs automatically with subsampling depending on the quality you choose (if I remember it correctly 4:2:0 with quality <=6 or 50% with save for web, 4:4:4 above).

this is true. but at this point you don't care much about quality so you won't benefit from having no subsampling anyway. I don't remember the case when had so save jpeg with something other than high quality. In fact most of the time I use 10-12. Most of the images I save have gradients of some sort.
The point is that after I saved gradient in webp with max quality, cause of subsampling or some other thing it looks bad compared to jpeg.

This is exactly my thoughts too:

Originally Posted by Glenn Maynard@google groops

don't believe a niche image format like this will ever be widely-adopted. I hope it isn't, because adoption of a niche format designed solely to save bandwidth will slow the creation and adoption of the *real* successor to JPEG.

[dapperdan]

Originally Posted by mandarinka

My results were much more modest (webp lossless filesizes were pretty unimpressive). Try if you don't get better compression with lossless jpeg 2000 or jpeg-XR (the former achieves better ratios). Make sure though that your software uses the reversible transform and yuv conversions. Jasper (also used in imagemagick) apparently fails to.

WebP lossless is basically a different file format from WebP. It's much more PNG-like rather than a lossless version of JPEG.

So you'd expect it to do better on typical web icon type images rather than photos (and it does: http://extrememoderate.wordpress.com/2011/11/28/a-web-centric-image-compression-benchmark/).

This is a fairly direct consequence of WebP in general being designed around making web images smaller and faster. If you've got a photorealistic image then the fastest, most efficient way to deliver it in a webpage is using lossy compression.

[IgorC]

Originally Posted by Keiyakusha

Even on official WebP presentation page all images converted to png for display. What joke is that?

Provide a link where only .png were posted.

[Keiyakusha]

Originally Posted by IgorC

Provide a link where only .png were posted.

I never said that "only png were posted". webm is there too, one of them even embedded in one of my posts. Here is a link for you, but I suggest to re-read what I said about png and browsers support.

[filler56789]

^

Hummm, I use Opera 11.64, and it supports only "normal" WebP. Lossless and lossy-with-alpha-channel get replaced with a placeholder : - /

And no, "full WebP support" is not a good reason for switching to Chrome/Chromium.

[vivan]

Originally Posted by filler56789

Hummm, I use Opera 11.64

The lastest Opera is 12.15. And it supports both webp lossless and lossy with alpha.

[filler56789]

[OT]
Opera 12.xx dropped the "Windows Native Skin"
[/OT]

[benwaggoner]

Originally Posted by tuqueque

@mzso... Boy, WebP is not only superior to JPEG in 95% of cases, but in Lossless mode is superior to PNG in 99% of cases... I'm already switching all of my "raw" lossless data to WebP and have saved around half the space. And considering that I work with PNG sequence animations, that's GBs of information I've saved!

I think there's a lost opportunity to use H.264 High Profile IDR frames as a still image format. Efficiency is better than WebP, and we'd get hardware decode support on most devices.

HEVC promises to be a very good still image format, particularly at high resolutions.

Getting support for arbitrary bit depths and channels would be great for all these formats, of course, so they could be used for RAW Bayer pattern and that sort of thing. I wish JPEG XR had taken off...

[pieter3d]

Originally Posted by benwaggoner

HEVC promises to be a very good still image format, particularly at high resolutions.

HEVC even has a special profile just for still images.

[pieter3d]

Here's what I know about VP9 coding tools.

64x64 "superblocks". Each divisible vertically, horzontally or quad. Quad divisions are recursive, so you can have any shape NxM, where N and M are 2^(2..6), and never more than 1 power apart (e.g. 4x4, 16x32, 8x4, 64x64 ... etc).
128x128 support is on the table, but undecided.

4 supported transforms: 4x4, 8x8, 16x16, 32x32. Yes this is the same as HEVC.

10 intra modes at any square size. 6 of these are directional.

1/8th pel motion comp, 8-tap filters. 3 filters (normal, blur, sharpen), selectable per prediction unit or frame.

3 ref pics per pic, but up to 8 pics to hold on to.

Discrete sine transform for intra blocks.

A second loop filter (analogous to SAO in HEVC) is under experimentation.

Bitstreams are still "data paritioned" like vp8, which is a pain for HW designs.

Entropy coding is similar to vp8.

Support for tiles, like HEVC, for parallel decode. Mandated above 2k resolution.

Support for lossless coding, as well as 4:4:4 with alpha.

The picture is still single-loop decodable to make HW implementations simpler.

[tuqueque]

http://www.androidauthority.com/webm...outube-209088/ (Let's try to keep this thread WebM-wise and not another H.26x one.)

[GTPVHD]

http://i.imgur.com/cqIQh6y.jpg

From Google I/O 2013.

[kypec]

Originally Posted by GTPVHD

http://i.imgur.com/cqIQh6y.jpg
From Google I/O 2013.

Awesome comparison! (Not!) If only they published what settings were used for H.264: Profile, Level, CABAC/CAVLC, ref-frames, B-frames, frame resolution, rate etc. etc.
Otherwise it's plain marketing rubbish, nothing more.

[nevcairiel]

It was a quick video show in a keynote, of course its marketing. There is no time in such an event for technical details.

[Keiyakusha]

Originally Posted by kypec

Awesome comparison! (Not!):mad

They won't benefit from showing something like this Looks like On2 guys taught them how to do the marketing.

[dapperdan]

The had a VP9-focused session directly after the keynote:

https://developers.google.com/events/io/sessions/325741299

But it's not one of the sessions being broadcast/recorded.

I assume they'll post the slides though.

edit: a news story based on the presentation

http://news.cnet.com/8301-1023_3-57584706-93/google-urges-fast-adoption-of-vp9-video-compression/

Best bits:

They claim VP9 saves ~50% bandwith compared with H.264 and is ~1% worse than H.265

There's a VP9 test channel:

http://www.youtube.com/user/WebMVP9/videos

And apparently you can turn on VP9 in Chrome today to view them.

[dapperdan]

The actual bit of the keynote when they talk about VP9 vs H.264

http://youtu.be/9pmPa_KxsAM?t=1h11m47s

They repeat the ~50% figure given above, the slightly larger figure is only for that particular video.

[cogman]

Originally Posted by dapperdan

The had a VP9-focused session directly after the keynote:

https://developers.google.com/events/io/sessions/325741299

But it's not one of the sessions being broadcast/recorded.

I assume they'll post the slides though.

edit: a news story based on the presentation

http://news.cnet.com/8301-1023_3-57584706-93/google-urges-fast-adoption-of-vp9-video-compression/

Best bits:

They claim VP9 saves ~50% bandwith compared with H.264 and is ~1% worse than H.265

There's a VP9 test channel:

http://www.youtube.com/user/WebMVP9/videos

And apparently you can turn on VP9 in Chrome today to view them.

http://www.youtube.com/watch?v=thtvQi6fEug&list=PLElsSrqQYMoHPY5k1l-X49OkW6Gz-9_zP&index=1

Ummm... wow, that is pretty terrible quality.

Grass is just one giant blob. Everything is VERY blocky, reminds me of the old xVid days. It is just really, well, terrible.

IDK, maybe the source video was this bad, but man, if they were trying to sell VP9 this video does them no favors.

[mk.2]

Originally Posted by Keiyakusha

Wikipedia says that WebP only supports 8-bit YUV 4:2:0 Is it true?

I think it's true.

I just tried converting Dark Shikari's avatar to WebP with XnView.

Here is the result:    This is original:

And that was converted with the best quality setting which suppose to be lossless. I also tried converting with WebP plugin for Photoshop. It produced the same result with slightly smaller filesize.

[vivan]

That just means that plugin doesn't support 4:4:4...