Google VP9 "Next Generation Open Video" information posted

[zerowalker]

Why can't any HW decoder do h264 10bit, why! ;(
Awesome that at least HEVC gets that.

[Nintendo Maniac 64]

Why can't any HW decoder do h264 10bit, why! ;(

I would guess it's because 10bit AVC is pretty much only ever used for anime bootlegs and demo videos and therefore is arguably too niche to spend resources on.

DISCLAIMER: I too would like 10bit AVC hardware decoding, but I never actually expected it to happen.

[mandarinka]

I think another issue is that the format is not covered by IP decoding blocks you can readily license. Or if it is available, the licensing costs are too high because it is considered pro/enterprise (because of AVC-intra).

Which means you have to do it on your own and it won't integrate with your already present circuitry, so it will be excess transistors, space, will need more complicated interfacing... it probably looks like too much pain to the bigger SoC vendors that have money for such stuff.

----------------------------

Edit:
Looks like the VP9 support is opencl/shader based GPU acceleration. It will be available to the Bristol Rdige APUs too (A8/A10/A12/FX).
http://images.anandtech.com/doci/10362/3%20-%20Stoney%20Ridge_575px.jpg
http://images.anandtech.com/galleries/4890/4%20-%20VP9%20HEVC_575px.jpg
http://images.anandtech.com/galleries/4890/6%20-%20A12_575px.jpg

Edit2:
http://www.amd.com/en-us/products/processors/laptop-processors#
This says that the quadcores are going to support 1080p VP9 ("FHD"), the dualcores (Stoney Ridge) only 720P it seems ("HD"). The latter is not so useful I guess.

[mandarinka]

I very much doubt the world will see 10bit vp9 outside of test sequences, unless hardware support becomes universal before AV1 shows up. I'd love it, but at this point it'll probably be in the same situation 10bit AVC was.

I was surprised too, but Kaby Lake, aka the next gen Intel CPUs launched this year (probably), are going to support hardware decoding of 10bit VP9.

Sources:
https://software.intel.com/sites/default/files/mediasdk_release_notes.pdf

Probably just profiles 0 and 2, not 1 and 3, although the API apparently aknowledges their existence.

This is probably due to 1) youtube planning HDR streaming this way, and 2) google/on2 probably provides IP that chipmakers can use to implement it, while such blocks aren't probably accessible enough for High 10 AVC.

[zerowalker]

Wait, is Youtube gonna use VP9 10bit?
Or am i reading it wrong, can't tell if you are debunking 10bit or not xD.

I would love to see it though, it would do wonders on certain games (Dark ones like Amnesia).

[NikosD]

Are there any VP9 encoders supporting 10 bit encoding ?

I'm trying to encode an uncompressed 10 bit stream to VP9 10bit.

[nevcairiel]

There is only one publically available encoder, libvpx. I don't know about the vpx commandline encoder, but at least using libvpx through ffmpeg allows encoding 10-bit if the libvpx it links to is build with high-bitdepth support.

[Selur]

..don't know about the vpx commandline encoder,..
vpxenc also support 10bit for quite some time now,...

[NikosD]

Thank you both

[mandarinka]

VP9 profile 2 support in ARM's IP block: http://www.anandtech.com/show/10428/arm-announces-mali-egil-video-processor

(Of course, shipping of chips with such hardware will take time, think 1-2 years depending on how long lead did close partners get to license it pre-announcement).

[Nintendo Maniac 64]

I'm more interested in the fact that the above article mentioned encode support and not just decode.

[mandarinka]

I'm more interested in the fact that the above article mentioned encode support and not just decode.

HEVC 10 encode is also a thing (supposedly for streaming HDR games...), so probably feature parity is the motivation. Or more checkboxes?

I didn't check VP9's 10-bit effectiveness gains, but I'm not sure they are that important, since we are talking mobile hardware encoding, which means shit quality from the start.

[Zetti]

Does there exist a freeware VP9 encoder??

[LigH]

How much more free do you want vpxenc or ffmpeg to become?!

[Zetti]

Before i wrote the post tried i search on google and did not find some useful results.

But i will try to search more on vpxenc and ffmpeg for VP9 encoding.

[LigH]

You should have read this thread instead, I guess... ;)

But vpxenc and ffmpeg are CLI encoders. You may also want to look for GUIs using them and providing an easier use.

[Selur]

As a gui, Hybrid supports VP8/9 encoding through vpxenc. (always happy to get some feedback regarding vp8/9 since I dropped using it myself, since a. feedback from devs took normally multiple month b. encoding was slow)

[Jamaika]

You can see that google really hiding from novelties.
As for ffmpeg it isn't possible to determine what version is vpxenc. I suppose that is the first version 1.5.0.
About the latest testing codecs AOM can forget because there are no players. Before you compile codec yourself. The joke lies in the fact that the developers not indicate how they write functions, "config".
Depending on how you configure these files "vpx_config.h, vp8/9/10/av1_enums.h, vp8/9/10/av1_rtcd.h, vp8/9/10/av1_common.h" you have different results at the end.

Greetings to all amateur enthusiasts.

[Jamaika]

Elimination of the truss pixels codecs VPX/AOM at low bitrates.

Theme was already before. Maybe a little interesting, because who uses VPX codec. However, I will return to him. I'll write how you can minimize pixelation.
VPX codec has VBR mode and it's set by default. Should change to the CRB mode {end-usage=cbr}. To increase the quality of the video frame it should be increased the percentage of the average per-frame bitrate {gf-cbr-boost}. The VP9 codec to 200% and AV1 codec to 300%. Does not affect the truss appearing function of {lag-in-frames}.

[wiak]

Polaris from amd has VP9 4k30 support it also has 10-bit HEVC
https://awesome.nwgat.ninja/rx480/Polaris-Tech-Day-Architecture-legally-approved/ef863980c6d742d6be8285356a039ac7-39.png
https://awesome.nwgat.ninja/rx480/Polaris-Tech-Day-Architecture-legally-approved/ef863980c6d742d6be8285356a039ac7-02.png

FYI if you want to read more on the Polaris chips aka Radeon RX 480/470 and 460
https://awesome.nwgat.ninja/rx480/

dunno about 10-bit VP9, perhaps the HW IP info section on the vp9 homepage has more info on what the open hardware design support

[Jamaika]

Does anyone know anything more about the novelties libvpx?
https://groups.google.com/a/webmproject.org/forum/#!topicsearch/webrtc/codec-devel/lo7VYVWrDFI
VP10 is still a work in progress and will be removed from the release v1.6.0
but significant changes have been made to both VP10 and AM1 and we
expect that to continue.

We would like to cut a release candidate within the next week or two.
What is a codec AM1? Maybe this is a typo. :confused:

[Nintendo Maniac 64]

What is a codec AM1? Maybe this is a typo. :confused:

Probably a typo and is referring to AV1, because what he describes is exactly what's going on with AV1.

Either that or it was before "AV1" was the agreed upon name and maybe it was known as AM1 internally or something.

[Jamaika]

Release v1.6.0 Khaki Campbell Duck (https://github.com/webmproject/libvpx/commit/042572177b5c58404836fce3fc221fd077dcd896)

This release improves upon the VP9 encoder and speeds up the encoding and
decoding processes.

- Upgrading:
This release is ABI incompatible with 1.5.0 due to a new 'color_range' enum
in vpx_image and some minor changes to the VP8_COMP structure.

The default key frame interval for VP9 has changed from 128 to 9999.

- Enhancement:
A core focus has been performance for low end Intel processors. SSSE3
instructions such as 'pshufb' have been avoided and instructions have been
reordered to better accommodate the more constrained pipelines.

As a result, devices based on Celeron processors have seen substantial
decoding improvements. From Indian Runner Duck to Javan Whistling Duck,
decoding speed improved between 10 and 30%. Between Javan Whistling Duck
and Khaki Campbell Duck, it improved another 10 to 15%.

While Celeron benefited most, Core-i5 also improved 5% and 10% between the
respective releases.

Realtime performance for WebRTC for both speed and quality has received a
lot of attention.

- Bug Fixes:
A number of fuzzing issues, found variously by Mozilla, Chromium and others,
have been fixed and we strongly recommend updating.

Delete codec VP10 with nextgen
Codec VP10 is in nextgenv2

[LigH]

"Khaki Campbell Duck" ~ (x)kcd :D

[mzso]

The default key frame interval for VP9 has changed from 128 to 9999.

Marvellous. I've been waiting for a long time for masses of unseekable junk videos.

[Motenai Yoda]

"Khaki Campbell Duck" ~ (x)kcd :D

https://imgs.xkcd.com/comics/standards.png

[Quikee]

The default key frame interval for VP9 has changed from 128 to 9999.

This is actually a mistake in the release notes. The change was from 9999 to 128.

[Jamaika]

Impossible. Libvpx v1.5.0 has keyframe interval 128. There are holiday so there is no upgrade ffmpeg order to check this.

[Quikee]

Impossible. Libvpx v1.5.0 has keyframe interval 128. There are holiday so there is no upgrade ffmpeg order to check this.

https://groups.google.com/a/webmproject.org/forum/#!topic/codec-devel/wlBMewAEBcg

and the exact commit that changed it to 128: https://chromium.googlesource.com/webm/libvpx/+/14828e756f92175c1a3c097c1048c01bbfba43a9%5E%21/vp9/vp9_cx_iface.c

[mandarinka]

14nm Radeons (Polaris) reportedly have VP9 decoding hardware (8bit only, likely) in addition to (10bit) HEVC.

The Stoney Ridge APU (E2-9010, A6-9210, A9-9410) according to some sources has VP9 decoding, according to some it does not (but maybe there could be a hybrid decoder available?). It does have 10-bit HEVC decoding though, that bit is confirmed.

The announced VP9 decoding on Polaris GPUs is reportedly really hybrid (shader-based GPU acceleration), if somebody had doubts.

[blurred]

Delete codec VP10 with nextgen
Codec VP10 is in nextgenv2[/I]
It seems two are being separately developed:
https://aomedia.googlesource.com/aom/
https://chromium.googlesource.com/webm/libvpx/+log/nextgenv2
They seem quite different - the first more Mozilla-Daala related, the second Google VP10.

Which one will become AV1?
Will VP10 be separately released?

[easyfab]

nextgenv2 is vp10 branch ( only google )
https://aomedia.googlesource.com/aom/ is AV1 branch ( with all aom members commits )

on chromium.googlesource.com the old AV1 branch was https://chromium.googlesource.com/webm/aom wich become aomedia.googlesource.com/aom/

[Jamaika]

Codec vp9 v1.6.0 in ffmpeg
Default GOP=128

[nakTT]

Sorry if this is not the right place to ask.

Any of you guys know any free software with extensive settings GUI like MeGUI, Handbrake etc that can do VP9 encoding?

I have been doing many encoding with MeGUI with x264 and x265 but have yet to get my hands on an application that can do VP9 conversion.

Thank you in advance.

[LigH]

Try Hybrid by Selur.

[nakTT]

Try Hybrid by Selur.
Thanks for the suggestion.

[Motenai Yoda]

Someone knows if the new Google Duo video-calling app uses vp9+opus? as it's said it can dynamically reduce the resolution, this sound very like a vp8/9/10 feature

[leonccyiu]

According to a PcPartpicker user who has the GTX1060, if you use Microsoft Edge, it implements hardware acceleration for VP9 on YouTube, he checked this by looking at stats for Nerds and CPU usage while watching 4k60. He also said that Neither Chrome, Firefox or Opera implemented VP9 hardware acceleration.

[wiak]

meybe edge supports Media Foundation and nvidia has a vp9 enabled Media Foundation codec?

[sneaker_ger]

MS explained it some time ago: https://blogs.windows.com/msedgedev/2016/04/18/webm-vp9-and-opus-support-in-microsoft-edge/

[leonccyiu]

it'd be nice if Chrome had dxva hardware acceleration for vp9 as google are the ones most aggressively pushing it. Hopefully it'll arrive soon.

[benwaggoner]

it'd be nice if Chrome had dxva hardware acceleration for vp9 as google are the ones most aggressively pushing it. Hopefully it'll arrive soon.


DXVA requires a driver-level codec implementation, typically GPU driver falling back to OS.

Or do you mean DirectComputer/OpenCL or something like that?


Sent from my iPad using Tapatalk

[huhn]

there is full DXVA VP9 support for a while now.

all pascal GPU are supporting this and it should be the same with polaris GPU.
even my "old" 960 maxwell GPU has a fixed function VP9 decoder.

DXVA VP9 support in web browser isn't a dream. it should happen at some time in the feature.

lavfilter is supporting it for over 6 month or something like that.

[NikosD]

Edge already supports it.

But it is for me a huge mystery, how on Earth the company (google) that develops both Chrome and VP9 hasn't managed yet to make HW acceleration feasible in its own browser that forces VP9 codec in Youtube, while Microsoft the other opponent has already done this in its own browser that doesn't forces the use of VP9.

Crazy.

[Blue_MiSfit]

It works in Chrome Canary ;)

[NikosD]

It works in Chrome Canary ;)
And how long is going to take approximately to go through the whole workflow, from Canary to dev from dev to beta and from beta to stable ?

[GTPVHD]

https://www.microsoft.com/en-us/download/details.aspx?id=49188

https://download.microsoft.com/download/9/2/A/92A4E198-67E0-4ABD-9DB7-635D711C2752/DXVA_VPx.pdf

August 2016
This document contains a specification for support of VP8 and VP9 video decoding (according to the available VP8 and VP9 source code and public documents) within the Microsoft Windows DirectX Video Acceleration (DXVA) API/DDI context.

This specification includes support of the VP8 coding format, VP9 Profile 0 (for 8 bit 4:2:0 video) and VP9 Profile 2 in 10 bit mode.

[x265_Project]

Netflix has posted on their Tech Blog, notifying people about a large study they have completed comparing x264, x265 and VP9 (libvpx). The results will be published Wednesday morning.
http://techblog.netflix.com/2016/08/a-large-scale-comparison-of-x264-x265.html

We ran a large-scale comparison of x264, x265 and libvpx to see for ourselves whether this 50% bandwidth improvement is applicable to our use case. Most codec comparisons in the past focused on evaluating what can be achieved by the bitstream syntax (using the reference software), applied settings that do not fully reflect our encoding scenario, or only covered a limited set of videos. Our goal was to assess what can be achieved by encoding with practical codecs that can be deployed to a production pipeline, on the Netflix catalog of movies and TV shows, with encoding parameters that are useful to a streaming service. We sampled 5000 12-second clips from our catalog, covering a wide range of genres and signal characteristics. With 3 codecs, 2 configurations, 3 resolutions (480p, 720p and 1080p) and 8 quality levels per configuration-resolution pair, we generated more than 200 million encoded frames. We applied six quality metrics - PSNR, PSNRMSE, SSIM, MS-SSIM, VIF and VMAF - resulting in more than half a million bitrate-quality curves. This encoding work required significant compute capacity. However, our cloud-based encoding infrastructure, which leverages unused Netflix-reserved AWS web servers dynamically, enabled us to complete the experiments in just a few weeks.

What did we learn?
Here’s a snapshot: x265 and libvpx demonstrate superior compression performance compared to x264, with bitrate savings reaching up to 50% especially at the higher resolutions. x265 outperforms libvpx for almost all resolutions and quality metrics, but the performance gap narrows (or even reverses) at 1080p.

Want to know more?
We will present our methodology and results this coming Wednesday, August 31, 8:00 am PDT at the SPIE Applications of Digital Image Processing conference, Session 7: Royalty-free Video. We will stream the whole session live on Periscope and YouTube: follow Anne for notifications or come back to this page for links to the live streams. This session will feature other interesting technical work from leaders in the field of Royalty-Free Video. We will also follow-up with a more detailed tech blog post and extend the results to include 4K encodes.

[mzso]

Netflix has posted on their Tech Blog, notifying people about a large study they have completed comparing x264, x265 and VP9 (libvpx). The results will be published Wednesday morning.
http://techblog.netflix.com/2016/08/a-large-scale-comparison-of-x264-x265.html

x265 outperforms libvpx for almost all resolutions and quality metrics, but the performance gap narrows (or even reverses) at 1080p.
This is weird... So it's worse below and above 1080p, but better or equivalent at 1080p?

[IgorC]

I wonder why MSU hasn't included VP9 this year http://compression.ru/video/codec_comparison/hevc_2016/