Versatile Video Coding (VVC) / H.266: HEVC successor

[Gravitator]

You can download > Test_VVC_AV1_x265
AV1 gives clearer textures on dark areas (lines and ship skin).
VVC:

-s 1920x1080 --framerate 23.976 --preset slower -q 29

AV1:

--cpu-used=0 --target-bitrate=1000 --max-partition-size=64 --sb-size=64 --input-bit-depth=10 --bit-depth=10

x265:

--output-depth 10 --profile main10 --qg-size 64 --opt-cu-delta-qp --tu-intra-depth 4 --tu-inter-depth 4 --hme --hme-search star,star,star --subme 4 --rect --amp --max-merge 2 --no-early-skip --rskip 0 --tskip --opt-ref-list-length-pps --bframes 6 --ref 4 --rc-lookahead 150 --bitrate 1100 --crf-min 0.00 --crf-max 0.00 --qpfile GENERATED_QP_FILE --limit-refs 0 --rd 5 --psy-rd 1.00 --rdoq-level 1 --psy-rdoq 10.00 --aq-strength 2 --sao-non-deblock --limit-sao --psnr --range limited --colormatrix bt709

[benwaggoner]

Originally Posted by birdie

Is this a question or a statement? If the latter it's very much debatable because from what I've seen VVC is much much better than AV1 while being a lot less computationally expensive.

Do you mean a lot less computationally expensive for encode or decode?

[birdie]

Originally Posted by benwaggoner

Do you mean a lot less computationally expensive for encode or decode?

Both.

[birdie]

Originally Posted by Gravitator

AV1 gives clearer textures on dark areas (lines and ship skin).

x265 has seen many years of refinements and improvements. x266 hasn't even been released yet. We are talking about proof of concept encoders.

Also, I've no idea how to play VVC.vvc. Please just upload PNGs to imgsli or something.

[Gravitator]

Originally Posted by birdie

Also, I've no idea how to play VVC.vvc.

A temporary solution...
Convert to Ut Video/lossless video format:

vvdecapp.exe -b c:\VVC.vvc -o c:\VVC.yuv && ffmpeg -video_size 1920x1080 -framerate 23976/1001 -pix_fmt yuv420p10le -i c:\VVC.yuv -c:v -acodec -vcodec utvideo -color_primaries 1 -color_trc 1 -colorspace 1 с:\VVC.mkv && del c:\VVC.yuv

Play it via FFplay:

vvdecapp.exe -b с:\VVC.vvc -o - | ffplay -f rawvideo -s 1920x1080 -pix_fmt yuv420p10le -

FFplay hot key control:

- q / ESC - exit
- f - full screen mode
- f / Space - pause
- m - on / off sound
- 9 / / - volume -
- 0 / * - volume +
- s - frame by frame
Rewind:
- left / right arrows - rewind 10 seconds
- up / down arrows - rewind 1 second
- Page up / Page down - rewind to the previous / next part or by 10 minutes

[Jamaika]

Originally Posted by birdie

x265 has seen many years of refinements and improvements. x266 hasn't even been released yet. We are talking about proof of concept encoders.

Also, I've no idea how to play VVC.vvc. Please just upload PNGs to imgsli or something.

For amateurs. VVC codec isn't created for converting bt709 PNG images to 10/12bit bt2020 fullrange images. Despite the HDR image, there is no desire to develop HDR for JPEG2000 or HEIF. JPEG XS is to have new possibilities, but no one has seen it. The new JPEG and VVC codecs are to have much greater data protection in the fight against piracy. Yes, not every file we open. Is VVC photo codec? Not. Can raw images from Canon apertures be converted to VVC. I failed. Converting CR2 to TIFF Adobe RGB is already lossy. In theory, it is possible. When we have converter suitable for yuv files and VVC codec with HDRtools functions. FFmpeg doesn't convert raw camera files to yuv bt2020. Problem with how to convert photos to yuv and then play back on TV. After all, TVs today hasn't VVC decoder and home monitors are 8bit bt709.

ffmpeg.exe -loglevel error -i 111.png -an -f rawvideo -frames 1 -vf scale=1280:720:in_color_matrix=bt709:in_range=fullut_color_matrix=bt709ut_range=full,format=yuv420p 111.yuv

EncoderApp.exe --SummaryVerboseness --InputFile=111.yuv --BitstreamFile=image.vvc --SourceWidth=1563 --SourceHeight=1558 --FrameRate=25.000 --InputBitDepth=8 --OutputBitDepth=10 --MSBExtendedBitDepth=8 --MatrixCoefficients=0 --InputColorPrimaries=-1 --LMCSSignalType=2 --ConformanceWindowMode=1 --FramesToBeEncoded=1 --HashME=1 --IBC=1 --DecodingRefreshType=1 --Profile=auto --InputSampleRange=1 --AspectRatioInfoPresent=1 --ChromaLocInfoPresent=1 --MaxCUWidth=16 --MaxCUHeight=16 --CTUSize=32 --MaxBTLumaISlice=32 --MaxBTChromaISlice=32 --MaxBTNonISlice=32 --MaxTTLumaISlice=32 --MaxTTChromaISlice=32 --MaxTTNonISlice=32 --CostMode=lossless --InputChromaFormat=444 --ChromaFormatIDC=444 --QP=0 --BDPCM=1 --ColorTransform=1 --VideoFullRange=1 --ChromaTS=1 --DepQuant=0 --LMCSEnable=0 --RDOQ=0 --RDOQTS=0 --SBT=0 --ISP=0 --MTS=0 --LFNST=0 --JointCbCr=0 --LoopFilterDisable=1 --SAO=0 --TransformSkip=1 --TransformSkipFast=1 --TransformSkipLog2MaxSize=5 --SAOLcuBoundary=0 --Log2MaxTbSize=5 --ALF=0 --CCALF=0 --BIO=0 --PROF=0 --InternalBitDepth=0 --IntraPeriod=1 --GOPSize=1 --SearchRange=64 --QpInValCb="17 22 34 42" --QpOutValCb="17 23 35 39" --BCW=0 --BcwFast=0 --BIO=0 --CIIP=0 --Geo=0 --AffineAmvr=0 --LMCSUpdateCtrl=1 --LMCSOffset=0 --DMVR=0 --SMVD=0 --PROF=0 --ISPFast=1 --FastMIP=1 --FastLFNST=1 --FastLocalDualTreeMode=0 --AffineAmvrEncOpt=0 --MmvdDisNum=8 --OnePictureOnlyConstraintFlag=1 --GciPresentFlag=1 --Level=15.5 --Tier=high --RateControl=0 --SubpicDecodedPictureHash=1 --MaxLayers=1 --CbQpOffset=1 --CrQpOffset=1 --TemporalSubsampleRatio=1 --LCTUFast=1 --TemporalFilter=0 --DualITree=1 --MinQTLumaISlice=8 --MinQTChromaISliceInChromaSamples=4 --MinQTNonISlice=8 --MaxMTTHierarchyDepth=3 --MaxMTTHierarchyDepthISliceL=3 --MaxMTTHierarchyDepthISliceC=3 --MMVD=1 --Affine=1 --SbTMVP=1 --MaxNumMergeCand=6 --LMChroma=1 --IMV=1 --MRL=1 --IBC=0 --AllowDisFracMMVD=1 --MIP=1 --PBIntraFast=1 --FastMrg=1 --AMaxBT=1 --HadamardME=1 --FEN=1 --FDM=1 --VerCollocatedChroma=1

[birdie]

Originally Posted by Jamaika

For amateurs.

I was talking about converting VVC to PNG which must be a lossless conversion

[Jamaika]

And here's the problem. Fraunhofer VVC Encoder isn't lossless. Is HDRtools enough for the image? Who knows that.

[birdie]

Originally Posted by Jamaika

And here's the problem. Fraunhofer VVC Encoder isn't lossless. Is HDRtools enough for the image? Who knows that.

I was talking about a lossy encode by Gravitator which would be easier for everyone to check without using console. Could you calm down a bit?

[Jamaika]

Lossy coding. Fraunhofer VVC only has basic functions. For me it is codec advertisement and an attempt to catch up with the competition. The codec is lagging behind in development.
Why is it free. I do not know. It is not the only company with vvc technology.
VVC has nothing to compare with x265 presets. The first HEVC Mainconcept codecs were also underdeveloped and the quality compared with x265 was deliberately poor. Whoever made movies in Sony Vegas at low bitrates knows what it looked like.
It changed when there was youtube.
Today Fraunhofer HEVC/VVC is trying to do something with these features and that's not just a challenge. It will take time to fine tune the presets and fix the bugs. In addition, it is trying to convert the codec from c++11 to c++20 with new features, which will also take some time.

[benwaggoner]

Originally Posted by birdie

I was talking about converting VVC to PNG which must be a lossless conversion

PNG is natively a full range RGB format. So you still have a color space and range conversion in the process. Normally it's fine going from 8-bit YUV to 8-bit RGB, with the inverse much more lossy. But it's still not lossless.

[benwaggoner]

Originally Posted by Jamaika

Lossy coding. Fraunhofer VVC only has basic functions. For me it is codec advertisement and an attempt to catch up with the competition. The codec is lagging behind in development.
Why is it free. I do not know. It is not the only company with vvc technology.
VVC has nothing to compare with x265 presets. The first HEVC Mainconcept codecs were also underdeveloped and the quality compared with x265 was deliberately poor. Whoever made movies in Sony Vegas at low bitrates knows what it looked like.
It changed when there was youtube.
Today Fraunhofer HEVC/VVC is trying to do something with these features and that's not just a challenge. It will take time to fine tune the presets and fix the bugs. In addition, it is trying to convert the codec from c++11 to c++20 with new features, which will also take some time.

It's quite normal for a variety of companies to offer their own encoder implementations in the early stage of a codec's market introduction. They've needed to build their own for internal testing, and might as well share them to promote market acceptance.

[birdie]

Originally Posted by benwaggoner

PNG is natively a full range RGB format. So you still have a color space and range conversion in the process. Normally it's fine going from 8-bit YUV to 8-bit RGB, with the inverse much more lossy. But it's still not lossless.

Slight changes in the color palette are barely perceivable however the loss of detail is what people will notice right away, so comparing codecs after converting their results to PNG is still valid. Besides, codecs work with the same color space which means converting them to PNG will affect them equally.

[hajj_3]

VTM 14.0 is out.

[Gravitator]

The reason for the encoder's refusal to assign an I-frame for frame №60 (scene change) ?
Themselves encode and check.

VVenC v1.1.0/VVdeC v1.1.2:

ffmpeg -i c:/raw_1080p30.mkv -pix_fmt yuv420p -f rawvideo - | vvencapp.exe -s 1920x1080 --framerate 30 --threads 2 --preset medium -q 30 -i - --output c:\test.vvc && vvdecapp.exe -b c:\test.vvc -o c:\VVC.yuv && ffmpeg -video_size 1920x1080 -framerate 30 -pix_fmt yuv420p10le -i c:\vvc.yuv -c:v -acodec -vcodec utvideo -color_primaries 1 -color_trc 1 -colorspace 1 c:\final.mkv && del c:\vvc.yuv

Download source video file (305.6MB) > raw_1080p30.mkv

[Balling]

Haha, wow. https://thenextweb.com/news/h266-vvc...ming-standards

First H.266 streamer.

[Balling]

Originally Posted by Gravitator

The reason for the encoder's refusal to assign an I-frame for frame №59 (scene change) ?
Themselves encode and check.

Not all I frames are key frames. /Facepalm

[Gravitator]

Originally Posted by Balling

Not all I frames are key frames. /Facepalm

I corrected it to the number 60. What's the difference? The encoder did not catch the change - it did not allocate the necessary bitrate. It turned out correctly on x265.

[Funky080900]

What do you mean VVenC did not allocate necessary bitrate? You encoded with constant QP (-q 30). If you want to use rate control, specify a bitrate E.g. -b 1000000. And there's no scene change detection in VVenC currently, so keyframes get placed in regular intervals.

[rwill]

Damn, guys...

I just remembered some test of VP3 where someone compared VP3 to DivX 3.11 and he set the keyframe period to 9999 sec or frames, can't remember, in VP3 and then continued to complain about quality and seek times. Now DivX 3.11 was (hacked?) to place keyframes on scenecuts but VP3 just did what it was told. Didnt serve it well to take user input literally.

@Gravitator: Would you complain about broadcast too ? Because most broadcast encoders do not insert I frames on scene change as well. They rather run with a large fixed open gop because they do not care about the screen content.