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. |
20th July 2021, 00:47 | #442 | Link |
Artem S. Tashkinov
Join Date: Dec 2006
Posts: 337
|
No idea for whom this could be useful for: Tencent cloud now supports H.266.
Last edited by birdie; 20th July 2021 at 00:50. |
21st July 2021, 09:38 | #443 | Link | ||
Broadcast Encoder
Join Date: Nov 2013
Location: Royal Borough of Kensington & Chelsea, UK
Posts: 2,883
|
Quote:
I'd like to see more companies getting ready to adopt the standard as they'll put the pressure on hardware manufacturers into actually supporting the decoding. Once we get there, it's done, ehehehehe In the meantime, I'm still eagerly waiting for a public release of x266, but I guess we'll still have to play with the reference encoder for a bit longer... Quote:
|
||
21st July 2021, 17:05 | #444 | Link | |
Registered User
Join Date: Mar 2009
Location: Russian Crimea
Posts: 335
|
Quote:
Last edited by VictorLS; 21st July 2021 at 17:29. |
|
21st July 2021, 17:17 | #445 | Link |
Registered User
Join Date: Jan 2019
Location: Canada
Posts: 570
|
It probably has to be demuxed into a raw VVC bitstream to feed to YUView, which then uses vvcdec to decode it into YUV.
I'm not sure what software supports the demuxing part though.
__________________
LG C2 OLED | GitHub Projects |
30th July 2021, 17:53 | #447 | Link | |
Broadcast Encoder
Join Date: Nov 2013
Location: Royal Borough of Kensington & Chelsea, UK
Posts: 2,883
|
Question number 1: Why is the Encoder using 01% of my 56c/112th Intel Xeon CPU and can we speed it up?
Question number 2: How can I get the same GOP size I would get with x262/x264/x265 when --ref 4 is set and nothing else is? Keyframes should be 24/250 by default, how can I get that here? I can't seem to get --IntraPeriod=1 --GOPSize=1 to work properly 'cause this way they're all intra, but as I change something, it complains about the GOP and it never works... Quote:
So far it's been 6 minutes and it still hasn't encoded a single frame... The performance is ridiculous and impossible to use on a daily basis. |
|
30th July 2021, 19:41 | #448 | Link | |||
Moderator
Join Date: Jan 2006
Location: Portland, OR
Posts: 4,750
|
Quote:
Quote:
Quote:
There is Fraunhofer's https://github.com/fraunhoferhhi/vvenc. It's explicitly meant to offer practical encoding times. x266 is also in development, although there haven't been any public releases yet. https://multicorewareinc.com/x266-a-...icoreware-inc/ It generally takes a couple of years from the bitstream standard being complete before stable and reasonably good quality/perf general-use encoders become broadly available. |
|||
30th July 2021, 19:58 | #449 | Link | ||
Registered User
Join Date: Aug 2019
Posts: 16
|
Quote:
https://github.com/fraunhoferhhi/vvenc/issues/28 Quote:
VVenC has presets and multithreading, so you probably want to use that one instead of VTM. Example: vvencapp -s 3840x2160 -r 30 -c yuv420 --internal-bitdepth 8 --preset faster --threads 16 --qp 32 -ip 256 -o VVC.266 -i input.yuv Last edited by Funky080900; 30th July 2021 at 20:06. |
||
30th July 2021, 22:15 | #450 | Link |
Registered User
Join Date: Jul 2015
Posts: 697
|
VTM has no frames IPB.
Try EVC xeve_app.exe -i "113.yuv" -o "output_10bit.evc" -v 2 -d 8 -m 4 -b 15 -I 256 --input-csp 1 --codec-bit-depth 10 --rc-type 1 --bitrate 3Mbps -w 1280 -h 720 -z 29 --frames 1000 --preset medium --level-idc 6.3 |
31st July 2021, 22:13 | #452 | Link |
Broadcast Encoder
Join Date: Nov 2013
Location: Royal Borough of Kensington & Chelsea, UK
Posts: 2,883
|
Thanks for the replies.
Of course I know that reference encoders are what they are, but since we have heard nothing for months and months from multicoreware about x266 (aside for the fact that they're working on it and that they won't make it open source until the right time comes), I was looking for a short time replacement to make some comparisons, but apparently the reference encoder is single threaded and most importantly it's Intra only, with no IPB GOP possible, which makes it unusable. @Jamaika... I'm gonna try with xeve_app.exe. Do you have a place where I can find the x64 Windows builds? Yep, it's something that has to be done in a frameserver like Avisynth etc, but this actually leads me to an interesting question: is interlace gone for good in H.266, or is it still gonna be around just to piss Ben? XD Last edited by FranceBB; 31st July 2021 at 22:19. |
31st July 2021, 23:53 | #453 | Link | |
Registered User
Join Date: Jan 2019
Location: Canada
Posts: 570
|
Quote:
Might be interesting to you. At least it's more recent than the announcement itself.
__________________
LG C2 OLED | GitHub Projects |
|
1st August 2021, 10:30 | #454 | Link |
Registered Developer
Join Date: Mar 2010
Location: Hamburg/Germany
Posts: 10,344
|
Seems like same level as support as HEVC, no specific interlace coding tools, but you can signal that a bitstream contains fields instead of frames - so you can encode interlaced that way.
__________________
LAV Filters - open source ffmpeg based media splitter and decoders |
1st August 2021, 11:09 | #455 | Link |
Registered User
Join Date: Jul 2015
Posts: 697
|
Question number 1: Why is the Encoder using 01% of my 56c/112th Intel Xeon CPU and can we speed it up?
If you used my codec it didn't have SIMD options SSE41, SSE42, AVX, AVX2, AVX512(3). EncoderApp hasn't std::thread function. VTM is only 32bit. No matter if it compiles to x86_64 Are my codecs working properly with multithreading in Windows 10? I don't know, because I don't know what are worth upgrading to gcc 12 in the mingw-std-thread add-on. https://github.com/Jamaika1/mingw_std_threads Code:
EncoderApp.exe --SummaryVerboseness --InputFile=test.yuv --BitstreamFile=video_2.vvc --SourceWidth=3840 --SourceHeight=2160 --FrameRate=30000/1001 --InputBitDepth=8 --OutputBitDepth=8 --MSBExtendedBitDep =8 --InternalBitDepth=0 --InputChromaFormat=420 --TargetBitrate=25000 --IntraPeriod=1 --GOPSize=1 --Profile=auto --Level=5.1 --Tier=wysoka --ChromaFormatIDC=420 --MatrixCoefficients=9 --InputColorPrimaries=1 --LMCSSignalType=2 --ConformanceWindowMode=1 --FramesToBeEncoded=6001 --HashME=1 --IBC=1 --DecodingRefreshType= 1 --InputSampleRange=1 --AspectRatioInfoPresent=1 --ChromaLocInfoPresent=1 --MaxCUWidth=32 --MaxCUHeight=32 --CTUSize=64 --MaxBTLumaISlice=64 --MaxBTChromaISlice=64 --MaxBTNonISlice=64 --MaxTTLumaISlice= 64 --MaxTTChromaISlice=64 --MaxTTNonISlice=64 --CostMode=stratna --BDPCM=0 --ColorTransform=0 --VideoFullRange=0 --ChromaTS=1 --DepQuant=1 --LMCSEnable=1 --RDOQ= 1 --RDOQTS=1 --SBT=1 --ISP=1 --MTS=1 --MTSIntraMaxCand=4 --MTSInterMaxCand=4 --LFNST=1 --JointCbCr=1 --SAO=1 --TransformSkip= 1 --TransformSkipFast=1 --TransformSkipLog2MaxSize=5 --SAOLcuBoundary=0 --ALF=1 --CCALF=1 --InputChromaFormat=422 --ChromaFormatIDC=422 --ColorTransform=1 --VideoFullRange=1--ChromaTS=1 --DepQuant=0 --LMCSEnable=0 --RDOQ=0 --RDOQTS=0 --SBT=0 --ISP=0 --MTS=3 --LFNST=0 --JointCbCr=0 --VerCollocatedChroma=1 --SAO=0 --TransformSkip=1 --TransformSkipFast=1 --TransformSkipLog2MaxSize=5 --SAOLcuBoundary=0 --Log2MaxTbSize=5 --ALF=0 --CCALF=0 --BIO=0 --PROF=0 --ZakresWyszukiwania=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=0 --RateControl=0 --SEIDecodedPictureHash=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 --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--Affine=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--Affine=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 Code:
# Type POC QPoffset QPOffsetModelOff QPOffsetModelScale CbQPoffset CrQPoffset QPfactor tcOffsetDiv2 betaOffsetDiv2 CbTcOffsetDiv2 CbBetaOffsetDiv2 CrTcOffsetDiv2 CrBetaOffsetDiv2 temporal_id #ref_pics_active_L0 #ref_pics_L0 reference_pictures_L0 #ref_pics_active_L1 #ref_pics_L1 reference_pictures_L1 Frame1: B 16 1 0.0 0.0 0 0 1.0 0 0 0 0 0 0 0 2 3 16 32 24 2 2 16 32 Frame2: B 8 1 -4.8848 0.2061 0 0 1.0 0 0 0 0 0 0 1 2 2 8 16 2 2 -8 8 Frame3: B 4 4 -5.7476 0.2286 0 0 1.0 0 0 0 0 0 0 2 2 2 4 12 2 2 -4 -12 Frame4: B 2 5 -5.90 0.2333 0 0 1.0 0 0 0 0 0 0 3 2 2 2 10 2 3 -2 -6 -14 Frame5: B 1 6 -7.1444 0.3 0 0 1.0 0 0 0 0 0 0 4 2 2 1 -1 2 4 -1 -3 -7 -15 Frame6: B 3 6 -7.1444 0.3 0 0 1.0 0 0 0 0 0 0 4 2 2 1 3 2 3 -1 -5 -13 Frame7: B 6 5 -5.90 0.2333 0 0 1.0 0 0 0 0 0 0 3 2 2 2 6 2 2 -2 -10 Frame8: B 5 6 -7.1444 0.3 0 0 1.0 0 0 0 0 0 0 4 2 2 1 5 2 3 -1 -3 -11 Frame9: B 7 6 -7.1444 0.3 0 0 1.0 0 0 0 0 0 0 4 2 3 1 3 7 2 2 -1 -9 Frame10: B 12 4 -5.7476 0.2286 0 0 1.0 0 0 0 0 0 0 2 2 2 4 12 2 2 -4 4 Frame11: B 10 5 -5.90 0.2333 0 0 1.0 0 0 0 0 0 0 3 2 2 2 10 2 2 -2 -6 Frame12: B 9 6 -7.1444 0.3 0 0 1.0 0 0 0 0 0 0 4 2 2 1 9 2 3 -1 -3 -7 Frame13: B 11 6 -7.1444 0.3 0 0 1.0 0 0 0 0 0 0 4 2 3 1 3 11 2 2 -1 -5 Frame14: B 14 5 -5.90 0.2333 0 0 1.0 0 0 0 0 0 0 3 2 3 2 6 14 2 2 -2 2 Frame15: B 13 6 -7.1444 0.3 0 0 1.0 0 0 0 0 0 0 4 2 3 1 5 13 2 2 -1 -3 Frame16: B 15 6 -7.1444 0.3 0 0 1.0 0 0 0 0 0 0 4 2 4 1 3 7 15 2 2 -1 1 Openvvc decoder. Everything would be OK, but I can't compile it. https://github.com/OpenVVC/OpenVVC/tree/dev Latest untested codecs in my garbage dump. https://www.sendspace.com/file/4ukanc Last edited by Jamaika; 1st August 2021 at 11:49. |
1st August 2021, 12:29 | #456 | Link | |
Registered User
Join Date: Aug 2019
Posts: 16
|
Quote:
|
|
1st August 2021, 12:59 | #457 | Link | |
Registered User
Join Date: Jul 2015
Posts: 697
|
Quote:
Code:
Parameter Check Error: GOP Size must be between 1 and 64 Parameter Check Error: GOP list auto config only supported GOP sizes: 1, 16, 32 cannot init the VVenC encoder VVEncoderApp.exe [error]: cannot create encoder, error -2 |
|
1st August 2021, 13:10 | #458 | Link |
Registered User
Join Date: Jul 2015
Posts: 697
|
If it is to be divisible by 1001 then you have to add it manually in the codec.
Code:
switch( vvencappCfg.m_FrameRate ) { case 23: temporalRate = 24000; temporalScale = 1001; break; case 29: temporalRate = 30000; temporalScale = 1001; break; case 59: temporalRate = 60000; temporalScale = 1001; break; default: break; } |
1st August 2021, 13:22 | #459 | Link |
Registered User
Join Date: Aug 2019
Posts: 16
|
Well it does say that supported GOP sizes are 1, 16 and 32. GOP size 1 is probably for low delay and all intra. Just leave GOP size at default.
It appears that FranceBB had a problem because he set --IntraPeriod=1 --GOPSize=1. Try --IntraPeriod=256 --GOPSize=32 instead. |
1st August 2021, 19:10 | #460 | Link |
Registered User
Join Date: Jul 2015
Posts: 697
|
Test codecs:
VVEncoderApp.exe -i "113.yuv" -o "output_10bit.vvc" -v 5 -t 4 -s 1280x720 -r 29 -c yuv420 -g 32 -ip 256 --internal-bitdepth 10 --bitrate 3000000 --passes 1 -f 1000 --preset medium --level 6.3 --tier high xeve_app.exe -i "113.yuv" -o "output_10abit.evc" -v 3 -d 8 -m 4 -w 1280 -h 720 -z 29 -b 15 -I 256 --input-csp 1 --codec-bit-depth 10 --rc-type 1 --bitrate 3Mbps --frames 1000 --preset medium --level-idc 6.3 --profile main EncoderApp.exe --SummaryVerboseness -c "encoder_randomaccess_vtm.cfg" --InputFile=113.yuv --BitstreamFile=video_2.vvc --SourceWidth=1280 --SourceHeight=720 --FrameRate=29 --InputBitDepth=8 --InternalBitDepth=12 --OutputBitDepth=10 --MSBExtendedBitDepth=8 --InputChromaFormat=420 --ChromaFormatIDC=420 --ConformanceWindowMode=1 --FramesToBeEncoded=1000 --LMCSSignalType=0 --Level=6.3 --BDPCM=1 --Tier=high --IBC=1 --OverscanInfoPresent=1 --AspectRatioInfoPresent=1 --ChromaLocInfoPresent=1 --Log2MaxTbSize=5 --VerCollocatedChroma=1 --DecodingRefreshType=1 VVEncoderApp is now faster than xeve, but incompatible with newer EncoderApp. Support for players is currently impossible because decoders don't have Y4M, i.e. they don't contain framerate, colormatrix, output bitdepth information. Standard xeve baseline profile is fast but has much poorer quality at low bitrates. Last edited by Jamaika; 1st August 2021 at 19:13. |
Thread Tools | Search this Thread |
Display Modes | |
|
|