Welcome to Doom9's Forum, THE in-place to be for everyone interested in DVD conversion.

Before you start posting please read the forum rules. By posting to this forum you agree to abide by the rules.

 

Go Back   Doom9's Forum > Video Encoding > VP9 and AV1

Reply
 
Thread Tools Search this Thread Display Modes
Old 24th February 2018, 01:40   #461  |  Link
Clare
Registered User
 
Join Date: Apr 2016
Posts: 61
objective2-slow https://people.xiph.org/~tdaede/sets/objective-2-slow/ has UHD content. Might be useful.
Clare is offline   Reply With Quote
Old 24th February 2018, 20:05   #462  |  Link
LigH
German doom9/Gleitz SuMo
 
LigH's Avatar
 
Join Date: Oct 2001
Location: Germany, rural Altmark
Posts: 6,753
New upload: AOM v0.1.0-8231-gcb4622dee
__________________

New German Gleitz board
MediaFire: x264 | x265 | VPx | AOM | Xvid
LigH is offline   Reply With Quote
Old 24th February 2018, 23:01   #463  |  Link
Tommy Carrot
Registered User
 
Tommy Carrot's Avatar
 
Join Date: Mar 2002
Posts: 863
Quote:
Originally Posted by Clare View Post
I'll be adding new data to my comparator with a AV1 snapshot from 20180222, next week or so. I've changed the encoder parameters for more speed so I need to recalculate the old snapshots to have comparable data.

Also I added the PIK image format.
Thanks for the comparisons.

The new AV1 blurs even more than the older ones. I know it's primarily a video encoder, so it's not really tuned for still images, but in its current form it's not really doing a good job as an image encoder.

However, PIK is really interesting. It basically behaves like a much more efficient JPEG. It has blocking and ringing artifacts, but preserves much more details than the other encoders. I sure prefer it over the artifactless, but sterile looking and blurry AV1 or HEVC images.
Tommy Carrot is offline   Reply With Quote
Old 26th February 2018, 02:11   #464  |  Link
Clare
Registered User
 
Join Date: Apr 2016
Posts: 61
Quote:
Originally Posted by Tommy Carrot View Post
Thanks for the comparisons.

The new AV1 blurs even more than the older ones. I know it's primarily a video encoder, so it's not really tuned for still images, but in its current form it's not really doing a good job as an image encoder.

However, PIK is really interesting. It basically behaves like a much more efficient JPEG. It has blocking and ringing artifacts, but preserves much more details than the other encoders. I sure prefer it over the artifactless, but sterile looking and blurry AV1 or HEVC images.
I've just stumble onto that on Github: https://aomediacodec.github.io/av1-avif/

Quote:
AV1 Still Image File Format (AVIF)

Overview

AVIF is a file format wrapping compressed images based on the Alliance for Open Media AV1 intra-frame encoding toolkit. AVIF supports High Dynamic Range (HDR) and wide color gamut (WCG) images as well as standard dynamic range (SDR). Only the intra-frame encoding toolkit is used in AVIF version 1.0. Using the intra-frame encoding mechanism from an existing video codec standard has a precedent in WebP: VP8, and HEIF: HEVC. This document describes at a high level a proposal on the structure of AVIF version 1.0.

The initial version of AVIF seeks to be simple, with just enough structure to allow the distribution of images based on the AV1 intra-frame coding toolset. At its core, AVIF 1.0 will allow for one or more images plus all supporting data needed to correctly reconstruct and display the images to be conveyed in a file. The ability to embed a thumbnail image will also be provided. An image sequence with suggested playback timing may be defined.
Clare is offline   Reply With Quote
Old 26th February 2018, 11:43   #465  |  Link
bstrobl
Registered User
 
Join Date: Jun 2016
Posts: 55
Quote:
Originally Posted by Clare View Post
I've just stumble onto that on Github: https://aomediacodec.github.io/av1-avif/
The part where it says:

An AVIF file should be a simplified and conformant version of an [HEIF] file.

would imply this is basically HEIF with AV1 as still image codec, which is a sensible choice.


Not a big fan of the ringing artefacts in PIK, those bother me quite a lot more.
bstrobl is offline   Reply With Quote
Old 26th February 2018, 13:19   #466  |  Link
mzso
Registered User
 
Join Date: Oct 2009
Posts: 930
Quote:
Originally Posted by Tommy Carrot View Post
Thanks for the comparisons.

The new AV1 blurs even more than the older ones. I know it's primarily a video encoder, so it's not really tuned for still images, but in its current form it's not really doing a good job as an image encoder.

However, PIK is really interesting. It basically behaves like a much more efficient JPEG. It has blocking and ringing artifacts, but preserves much more details than the other encoders. I sure prefer it over the artifactless, but sterile looking and blurry AV1 or HEVC images.
I wonder why they bother with HEIF/ISOBMFF when AV1 is already webm/matroska centric.
mzso is offline   Reply With Quote
Old 28th February 2018, 02:28   #467  |  Link
TD-Linux
Registered User
 
Join Date: Aug 2015
Posts: 34
Quote:
Originally Posted by Tommy Carrot View Post
The new AV1 blurs even more than the older ones. I know it's primarily a video encoder, so it's not really tuned for still images, but in its current form it's not really doing a good job as an image encoder.
If you want to experiment, you can build an encoder with -DCONFIG_DIST_8X8=1 and then specify --tune=cdef-dist or --tune=daala-dist. These will optimize away from blurring and towards ringing.
TD-Linux is offline   Reply With Quote
Old 4th March 2018, 12:09   #468  |  Link
bstrobl
Registered User
 
Join Date: Jun 2016
Posts: 55
Here is the buglist for the bitstream:

https://bugs.chromium.org/p/aomedia/...mary&cells=ids

Presumably once all of those have been fixed the bitstream can be frozen.


Looks like everyone is working on weekends to get this thing done.

Last edited by bstrobl; 4th March 2018 at 12:23.
bstrobl is offline   Reply With Quote
Old 4th March 2018, 15:11   #469  |  Link
hajj_3
Registered User
 
Join Date: Mar 2004
Posts: 1,120
Quote:
Originally Posted by bstrobl View Post
Here is the buglist for the bitstream:

https://bugs.chromium.org/p/aomedia/...mary&cells=ids

Presumably once all of those have been fixed the bitstream can be frozen.


Looks like everyone is working on weekends to get this thing done.
not sure, I have been following this list: https://bugs.chromium.org/p/aomedia/...um=100&start=0

It is down to 75 bugs, 2.5 weeks ago it was at 215 bugs so probably another 1-2 weeks or so until it is ratified.
hajj_3 is offline   Reply With Quote
Old 4th March 2018, 18:40   #470  |  Link
Clare
Registered User
 
Join Date: Apr 2016
Posts: 61
Quote:
Originally Posted by Clare View Post
I also plan to do an actual video comparaison based on 30 short clips, VMAF metrics, AV1, X264, X265, and VP9. Gonna take a long while as I haven't written a single line of code yet and the encoding itself will be long
Quote:
Originally Posted by Clare View Post
I plan to use objective1-fast https://people.xiph.org/~tdaede/sets/objective-1-fast/
It's a mix of 360p, 720p and 1080p. I don't have any 1440p or UHD content and I doubt my computer would be able to process it in a reasonable time. But I plan to release the Python scripts I use on Github so it will be usable on any dataset, like I did for images (https://github.com/WyohKnott/image-comparison-sources).
So it was easier to code than I thought, with few changes from the image comparison.

It's up there: https://wyohknott.github.io/video-formats-comparison/

I haven't added AV1 yet, I'm waiting for the bitstream freeze.

Let me know what you think would be good to add. I put most of the graph I thouqgt would be useful. I wish I could addsome examples videos to plày side by side but I don't know how to do that.
Clare is offline   Reply With Quote
Old 4th March 2018, 19:47   #471  |  Link
Jamaika
Registered User
 
Join Date: Jul 2015
Posts: 697
Code:
Each Y4M videos is exported to 4:2:0 10 bits Y4M with FFMPEG:

ffmpeg -y -i [input] -pix_fmt yuv420p10le -strict -1 [output]
Video compression
All videos are compressed over a range of qualities for each encoder:
•AV1:
◦between q=12 and q=60, with a step of 4:

aomenc --cpu-used=2 --tile-columns=4 --passes=2 --pass=1 --bit-depth=10 --input-bit-depth=10 --end-usage=q --cq-level=$q --fpf=[output].log -o [output] [input(Y4M_10bits)]

aomenc --cpu-used=2 --tile-columns=4 --passes=2 --pass=2 --bit-depth=10 --input-bit-depth=10 --end-usage=q --cq-level=$q --fpf=[output].log -o [output] [input(Y4M_10bits)]

•VP9:
◦between q=12 and q=60, with a step of 4:

vpxenc --tile-columns=4 --row-mt=1 --passes=2 --cpu-used=2 --bit-depth=10 --input-bit-depth=10 --profile=2 --end-usage=q --cq-level=$ -o [output] [input(Y4M_10bits)]

•x264:
◦between q=12 and q=40, with a step of 2:


x264 --profile high10 --preset slower --input-depth=10 --output-depth=10 --crf $q -o [output] [input(Y4M_10bits)]

•x265:
◦between q=12 and q=40, with a step of 2:


x265 --profile main10 --preset slower --input-depth=10 --output-depth=10 --crf $q -o [output] [input(Y4M_10bits)]
My ideas:
Code:
ffmpeg.exe -loglevel verbose -i "[input(your_colorspace).raw]" -an -f yuv4mpegpipe -vf scale=1920:1080:in_color_matrix=rgb:in_range=full:out_color_matrix=bt2020_ncl:out_range=full,format=yuv422p10le -strict -1 -
ffmpeg.exe -loglevel verbose -i "[input(your_colorspace).y4m]" -an -f yuv4mpegpipe -vf scale=1920:1080:in_color_matrix=S_gamut3:in_range=full:out_color_matrix=bt2020_ncl:out_range=full,format=yuv422p10le,lutyuv=val:val:val -strict -1 -
Code:
aomenc.exe --bit-depth=10 --input-bit-depth=10 --i4?? --codec=av1 --good --threads=? --cpu-used=? --profile=3 --drop-frame=(0 or 100 depends on fps) --end-usage=q --cq-level=48 --target-bitrate=0 --min-q=?(the problem bitrate of the first frame) 
--kf-max-dist=GOP --auto-alt-ref=? --frame-boost=? --aq-mode=0 --color-space=bt2020 --verbose --debug --pass=? --passes=? --output="xxx.webm" - (where is range full?)
Code:
x264.exe --demuxer y4m --input-csp i4?? --input-depth 10 --input-range pc --output-csp i4?? --threads ? --preset slow --tune grain --crf 28 --fps ??.??? --keyint 60 --nal-hrd none --vbv-bufsize 40000 --vbv-maxrate 40000(use for web or don't use) 
--colormatrix bt2020nc --colorprim bt2020 --transfer bt2020-10 --range pc --output "xxx.h264" -
Jamaika is offline   Reply With Quote
Old 5th March 2018, 08:07   #472  |  Link
o-l-a-v
Registered User
 
Join Date: Feb 2013
Posts: 13
Quote:
Originally Posted by Clare View Post
You are linking to PIKs github page for x265.

"x265: https://github.com/google/pik. The version used is built from GIT revision 3cf3839f82bb177c43449ab10792c184c4485d8b"
o-l-a-v is offline   Reply With Quote
Old 5th March 2018, 13:26   #473  |  Link
Tommy Carrot
Registered User
 
Tommy Carrot's Avatar
 
Join Date: Mar 2002
Posts: 863
Quote:
Originally Posted by Clare View Post
So it was easier to code than I thought, with few changes from the image comparison.

It's up there: https://wyohknott.github.io/video-formats-comparison/

I haven't added AV1 yet, I'm waiting for the bitstream freeze.

Let me know what you think would be good to add. I put most of the graph I thouqgt would be useful. I wish I could addsome examples videos to plày side by side but I don't know how to do that.
Nice.

I think XVC encoder could be interesting to see. I know it's a proprietary codec, but quality-wise, in my experience, it competes quite well with x265 and AV1.
Tommy Carrot is offline   Reply With Quote
Old 5th March 2018, 17:07   #474  |  Link
clsid
*****
 
Join Date: Feb 2005
Posts: 5,642
Does anyone have a link to a sample file that was encoded by a recent build?
clsid is offline   Reply With Quote
Old 5th March 2018, 23:33   #475  |  Link
LigH
German doom9/Gleitz SuMo
 
LigH's Avatar
 
Join Date: Oct 2001
Location: Germany, rural Altmark
Posts: 6,753
Just for you ... I created a "samples" subdirectory in my AOM MediaFire folder; it contains a WebM clip of "foreman" in PAL CIF resolution, encoded with AOM encoder v0.1.0-8300-g088217b2e (built Feb. 27, I hope that's recent enough) with '--cpu-used 8' (supposed to be a faster preset). Encoding took 3/4 hour on this aged PC with AMD Phenom-II (max. SSE2).

Tomorrow I could try a newer one with faster hardware.
__________________

New German Gleitz board
MediaFire: x264 | x265 | VPx | AOM | Xvid
LigH is offline   Reply With Quote
Old 6th March 2018, 16:13   #476  |  Link
LigH
German doom9/Gleitz SuMo
 
LigH's Avatar
 
Join Date: Oct 2001
Location: Germany, rural Altmark
Posts: 6,753
A few more uploads: AOM v0.1.0-8449-g6471e8bd7 and samples with "--cpu-used [8..4]" to compare speeds (encoded on an AMD FX-6300). Disillusioning...

Code:
Pass 2/2 frame  300/300   469809B   12528b/f  313206b/s  769973 ms (0.39 fps)
Pass 2/2 frame  300/300   460022B   12267b/f  306681b/s  841144 ms (0.36 fps)
Pass 2/2 frame  300/300   449454B   11985b/f  299636b/s  921242 ms (0.33 fps)
Pass 2/2 frame  300/300   448798B   11967b/f  299198b/s 1044732 ms (0.29 fps)
Pass 2/2 frame  300/300   448035B   11947b/f  298690b/s 1214326 ms (0.25 fps)
Reminding you, PAL CIF resolution (352×288). Due to the small video, only 1 thread was active.
__________________

New German Gleitz board
MediaFire: x264 | x265 | VPx | AOM | Xvid
LigH is offline   Reply With Quote
Old 6th March 2018, 18:30   #477  |  Link
hbbs
Registered User
 
Join Date: Nov 2015
Posts: 9
Sorry, I am fairly new to this.

But which player are you using to play this samples?

I have tried the latest vlc stable to no avail.

Couldn't play the codec "av01" (AOMedia's AV1 Video)
hbbs is offline   Reply With Quote
Old 6th March 2018, 22:06   #478  |  Link
clsid
*****
 
Join Date: Feb 2005
Posts: 5,642
Thanks LigH.

The AV1 format has not been finalized yet. Once that is done, you can expect players to quickly adopt it.
clsid is offline   Reply With Quote
Old 6th March 2018, 22:06   #479  |  Link
LigH
German doom9/Gleitz SuMo
 
LigH's Avatar
 
Join Date: Oct 2001
Location: Germany, rural Altmark
Posts: 6,753
There is probably no player available yet. You can only use aomdec to decode to Y4M and play this decoded raw video, AFAIK.
__________________

New German Gleitz board
MediaFire: x264 | x265 | VPx | AOM | Xvid
LigH is offline   Reply With Quote
Old 6th March 2018, 23:03   #480  |  Link
Tommy Carrot
Registered User
 
Tommy Carrot's Avatar
 
Join Date: Mar 2002
Posts: 863
Thanks for the new build.

Despite all the SIMD optimizations and other improvements, this build is actually even slower than build #8039. The quality seems visually very similar, the metrics as well. I was curious what AV1 is capable of with all the compression tools enabled, so i tried --cpu-used=0 on a 50 frames long, 720*288 video. It took almost 6 hours to encode.
Tommy Carrot is offline   Reply With Quote
Reply

Thread Tools Search this Thread
Search this Thread:

Advanced Search
Display Modes

Posting Rules
You may not post new threads
You may not post replies
You may not post attachments
You may not edit your posts

BB code is On
Smilies are On
[IMG] code is On
HTML code is Off

Forum Jump


All times are GMT +1. The time now is 16:43.


Powered by vBulletin® Version 3.8.11
Copyright ©2000 - 2024, vBulletin Solutions Inc.