LAV Filters - DirectShow Media Splitter and Decoders

[Marsu42]

Originally Posted by nevcairiel

LAV Filters 0.75
[CODE]
One caveat for hardware decoding, film grain synthesis may not work on Intel or AMD. Due to lack of hardware I did not have the ability to test. On NVIDIA 30-series, it works, however.

Is there a specific concern film grain synthesis might not work on non-Nvidia hardware, or is it simply due to lack of test hardware? Thanks!

[mts1]

Hi guys.

I have a question about VP9 hardware acceleration using Radeon WX 7100.
Is it possible to have it with LAV Filters?
If not, would it be possible to have it in the near future? Or this is dead end?
Perhaps you know any other way/filters to achieve that?

Thanks.

[huhn]

this is a polaris based card the VP9 decoder is hybrid which is often worse than just using the CPU and is currently not available at all there where only a could of old drivers where it worked in chrome.

i have not rechecked this information in a long time so things could be different but it should still be a dead end.

[levizx]

Originally Posted by nevcairiel

Its a one-time setup step to select the more capable GPU.

yes, it's more capable, but also much more power hungry. that's not good for a laptop and sometimes defeats the purpose of having hw decoder altogether.

[mts1]

Originally Posted by huhn

this is a polaris based card the VP9 decoder is hybrid which is often worse than just using the CPU and is currently not available at all there where only a could of old drivers where it worked in chrome.

i have not rechecked this information in a long time so things could be different but it should still be a dead end.

Got it.
Thanks.

[Sunspark]

I think I have located a video decoding bug, but I am not certain if the issue is LAV or another part of the chain.

A 10-bit P010 anime file is the source, being displayed in D3D9 exclusive on Windows 10.

DXVA2 Copy-Back - Works
DXVA2 Native - Video is split in two, left half is purple tinted, and right half is a solid block of green
Intel Quicksync - Works
DX11 Native - Image freezes
DX11 Copy-Back - Works

The issue with the DXVA2 and DX11 Native modes doesn't happen with non-P010 videos. I also checked and it was present in the previous release as well.

[wanezhiling]

Hi, lav hw decoding seems not work with this clip, always sw decoding.

anyone confirm?

[el Filou]

Originally Posted by Sunspark

A 10-bit P010 anime file is the source [...] The issue with the DXVA2 and DX11 Native modes doesn't happen with non-P010 videos.

P010 is normally an output format, what's the video format + encoding (level, profile, other coding parameters)? Maybe post the video section of MediaInfo.
What's the hardware and driver? Does this also happen with older/newer drivers?
What's the video renderer? Does it also happen with other renderers/other modes than D3D9 exclusive?

[Sunspark]

Originally Posted by el Filou

P010 is normally an output format, what's the video format + encoding (level, profile, other coding parameters)? Maybe post the video section of MediaInfo.
What's the hardware and driver? Does this also happen with older/newer drivers?
What's the video renderer? Does it also happen with other renderers/other modes than D3D9 exclusive?

Hardware is an Intel HD Graphics 6000 (Broadwell CPU). MadVR .92.17 is the video renderer. The image/chroma scaling I'm using with it is DXVA.

Mediainfo below. Looking at it, I didn't realize that it was HEVC.. Interesting! I guess that explains why my CPU was running higher than expected while it was playing since on this GPU for things like VP9 it runs in hybrid mode (both cpu and some functions in gpu).

Video
ID                             : 1
Format                         : HEVC
Format/Info                    : High Efficiency Video Coding
Format profile                 : Main 10@L4@Main
Codec ID                       : V_MPEGH/ISO/HEVC
Duration                       : 24 min 21 s
Bit rate                       : 3 078 kb/s
Width                          : 1 920 pixels
Height                         : 1 080 pixels
Display aspect ratio           : 16:9
Frame rate mode                : Constant
Frame rate                     : 23.976 (24000/1001) FPS
Color space                    : YUV
Chroma subsampling             : 4:2:0
Bit depth                      : 10 bits
Bits/(Pixel*Frame)             : 0.062
Stream size                    : 536 MiB (87%)
Writing library                : x265 3.3+2-gbe2d82093:[Windows][GCC 9.2.0][64 bit] 10bit
Encoding settings              : cpuid=1111039 / frame-threads=4 / numa-pools=24 / wpp / no-pmode / no-pme / no-psnr / no-ssim / log-level=2 / input-csp=1 / input-res=1920x1080 / interlace=0 / total-frames=35040 / level-idc=0 / high-tier=1 / uhd-bd=0 / ref=4 / no-allow-non-conformance / no-repeat-headers / annexb / no-aud / no-hrd / info / hash=0 / no-temporal-layers / open-gop / min-keyint=23 / keyint=250 / gop-lookahead=0 / bframes=8 / b-adapt=2 / b-pyramid / bframe-bias=0 / rc-lookahead=25 / lookahead-slices=4 / scenecut=40 / hist-scenecut=0 / radl=0 / no-splice / no-intra-refresh / ctu=64 / min-cu-size=8 / rect / no-amp / max-tu-size=32 / tu-inter-depth=1 / tu-intra-depth=1 / limit-tu=0 / rdoq-level=2 / dynamic-rd=0.00 / no-ssim-rd / signhide / no-tskip / nr-intra=0 / nr-inter=0 / no-constrained-intra / strong-intra-smoothing / max-merge=3 / limit-refs=3 / limit-modes / me=3 / subme=3 / merange=57 / temporal-mvp / no-frame-dup / no-hme / weightp / no-weightb / no-analyze-src-pics / deblock=1:1 / sao / no-sao-non-deblock / rd=4 / selective-sao=4 / no-early-skip / rskip / no-fast-intra / no-tskip-fast / no-cu-lossless / no-b-intra / no-splitrd-skip / rdpenalty=0 / psy-rd=1.00 / psy-rdoq=1.00 / no-rd-refine / no-lossless / cbqpoffs=0 / crqpoffs=0 / rc=crf / crf=18.0 / qcomp=0.60 / qpstep=4 / stats-write=0 / stats-read=0 / ipratio=1.40 / pbratio=1.30 / aq-mode=3 / aq-strength=0.80 / cutree / zone-count=0 / no-strict-cbr / qg-size=32 / no-rc-grain / qpmax=69 / qpmin=0 / no-const-vbv / sar=0 / overscan=0 / videoformat=5 / range=0 / colorprim=2 / transfer=2 / colormatrix=2 / chromaloc=0 / display-window=0 / cll=0,0 / min-luma=0 / max-luma=1023 / log2-max-poc-lsb=8 / vui-timing-info / vui-hrd-info / slices=1 / no-opt-qp-pps / no-opt-ref-list-length-pps / no-multi-pass-opt-rps / scenecut-bias=0.05 / hist-threshold=0.01 / no-opt-cu-delta-qp / no-aq-motion / no-hdr10 / no-hdr10-opt / no-dhdr10-opt / no-idr-recovery-sei / analysis-reuse-level=0 / analysis-save-reuse-level=0 / analysis-load-reuse-level=0 / scale-factor=0 / refine-intra=0 / refine-inter=0 / refine-mv=1 / refine-ctu-distortion=0 / no-limit-sao / ctu-info=0 / no-lowpass-dct / refine-analysis-type=0 / copy-pic=1 / max-ausize-factor=1.0 / no-dynamic-refine / no-single-sei / no-hevc-aq / no-svt / no-field / qp-adaptation-range=1.00 / no-scenecut-aware-qpconformance-window-offsets / right=0 / bottom=0
Default                        : Yes
Forced                         : No

[nevcairiel]

Originally Posted by wanezhiling

Hi, lav hw decoding seems not work with this clip, always sw decoding.

anyone confirm?

The video has too many reference frames. H.264 4K at Level 5.1 only supports up to 5, not 16 as the video uses.

[DMU]

Originally Posted by wanezhiling

anyone confirm?

CPU: AMD Ryzen 3 3200G with Radeon Vega Graphics
GPU: AMD Radeon(TM) Vega 8 Graphics
Decoder: LAV Video Decoder
Decoder Device: H264_VLD_NoFGT
Frames: 1798
FPS: 174.177 [152-182]
CPU Usage: -
GPU 3D Engine Usage: 15 [14-16] %
GPU Video Codec 0 Engine Usage: 93 [86-95] %

[wanezhiling]

Originally Posted by nevcairiel

The video has too many reference frames. H.264 4K at Level 5.1 only supports up to 5, not 16 as the video uses.

Got it, thanks.

Originally Posted by DMU

CPU: AMD Ryzen 3 3200G with Radeon Vega Graphics
GPU: AMD Radeon(TM) Vega 8 Graphics
Decoder: LAV Video Decoder
Decoder Device: H264_VLD_NoFGT
Frames: 1798
FPS: 174.177 [152-182]
CPU Usage: -
GPU 3D Engine Usage: 15 [14-16] %
GPU Video Codec 0 Engine Usage: 93 [86-95] %

As nevcairiel explained above, LAV won't decode such file in hw, so maybe something is wrong on your side?

[DMU]

I would be glad if you find something wrong on my side.
Pic.

[wanezhiling]

Originally Posted by DMU

I would be glad if you find something wrong on my side.
Pic.

Because you are using an old LAV build(0.74.1-92) which still supports such H.264 file in hw decoding.

Update your MPC-HC and your lav will lose that ability.

[el Filou]

Originally Posted by Sunspark

Hardware is an Intel HD Graphics 6000 (Broadwell CPU). [...] Looking at it, I didn't realize that it was HEVC.. Interesting!

Interesting indeed, because IIRC the HEVC decoder in Broadwell was hybrid (used GPU compute and not fixed function unit), so maybe that's where the difference comes from? I would think it's a driver problem but it's hard to know really. Does it give a corrupt image too when using standard EVR ?

[VictorLS]

Originally Posted by wanezhiling

Because you are using an old LAV build(0.74.1-92) which still supports such H.264 file in hw decoding.
Update your MPC-HC and your lav will lose that ability.

Even next 0.74.1-98 can't DXVA2 HW acceleration of your Girls.Generation.Oh.4in1.201002.HDTV.x264.2160p.120fps.DTSES.6.1ch.mkv but even newest 0.75.0-2 can old NVIDIA CUVID acceleration with about 100 MB more videomemory consumption (so about 700 MB of 1 GB of my GTX750v2 GM206 videocard) than DXVA2 native in 0.74.1-92 but NVIDIA CUVID can't decode all frames while audio stops (DXVA2 native in 0.74.1-92 and also Cyberlink, MainConcept and MPC-BE Video Decoder with DXVA can).

[nevcairiel]

Don't encode your videos out of spec if you want them to work better. If it actually uses 16 ref frames on 4K, then it would artifact decoding in hardware, I didn't add those checks out of nowhere, and this was on modern NVIDIA hardware too.
If it doesn't use all 16 but is flagged to contain 16, then the stream is just bad. As a general rule, 16 ref frames rarely offer a solid advantage, which is why it was reduced on HEVC again to max 8 active ref frames per frame as well.

Artifacts are the worst thing that can happen. I rather not decode some obscure out-of-spec video in hardware then risk artifacts. That has always been my stance on this.

[Sunspark]

Originally Posted by el Filou

Interesting indeed, because IIRC the HEVC decoder in Broadwell was hybrid (used GPU compute and not fixed function unit), so maybe that's where the difference comes from? I would think it's a driver problem but it's hard to know really. Does it give a corrupt image too when using standard EVR ?

No, I checked just now with EVR Custom Presenter, and it displayed fine in DXVA2 Native mode. Thanks for the suggestion to check that.

So based on this, it appears that the bug may be with MadVR after all which hasn't received a public update for a couple years now.

Update: I found a way to make it work in DXVA2 Native mode w/ MadVR for this type of video. If you uncheck "Present frames in advance" it will change it to backbuffers which is D3D9 Old Path. Frames in advance is D3D9 New Path. Not completely sure what the difference between Old and New is, but old works if you want to use Native instead of Copy-Back and even more curious, it very subtly appears to be smoother on panning motion than new.

[VictorLS]

Originally Posted by nevcairiel

Don't encode your videos out of spec if you want them to work better.

Most cases is playing enocoded by someone files (streams, i.e. from SAT) but not own encoded.

Originally Posted by nevcairiel

Artifacts are the worst thing that can happen. I rather not decode some obscure out-of-spec video in hardware then risk artifacts. That has always been my stance on this.

So you still live in ideal world and still don't want to give choice for people as v0lt in their modern MPC-VR with always turning on HLG processing or MPC-BE Video Decoder - just one version has Skip non I-frames (newer has Skip B-frames only isn't enough for 8K in my case)

I use 1.5.3(build4166) only - in case of LAV Video Decoder I'll have to use i.e. 0.74.1-92 in case of playing such Girls.Generation.Oh.4in1.201002.HDTV.x264.2160p.120fps.DTSES.6.1ch.mkv files but I prefer use one (desirable most modern) and most universal video decoder with various settings for all possible cases - if I see artifacts with DXVA2 I prefer just change acceleration to NVIDIA CUVID then install older version of LAV Video Decoder

[nevcairiel]

Originally Posted by VictorLS

So you still live in ideal world and still don't want to give choice for people

Luckily, I don't have to care what you think, and those options will never happen. You are free to not use my software if you don't like my decisions. I don't get anything from it.