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

[Funky080900]

ffmpeg.exe -i "AVS Script.avs" -strict -1 -an -f yuv4mpegpipe - | vvencapp.exe -i - --preset medium --format yuv420 --bitrate 25000 --profile auto --level 5.0 --hdr pq --passes 1 --output "VVC.h266"

pause

In VVenC bitrate is specified in bits per second rather than kilobits/second. So 25000 is just 25 kbps which is probably not what you wanted.

[FranceBB]

Originally Posted by Funky080900

In VVenC bitrate is specified in bits per second rather than kilobits/second. So 25000 is just 25 kbps which is probably not what you wanted.

Dammit, I figured it out when I saw the final file being as little as 14 MB and I was like: WUT?!
Now I'm gonna try again with 25000000 bits per second which should be 25000 kbits/s so 25 Mbit/s and after that I'll show you some nice SSIM/PSNR stats.

Originally Posted by ksec

Ok the last pic is just showing off

Eheheheh you got me xD

Originally Posted by benwaggoner

What kind of fps did you get?

It doesn't say in the cmd, but it was definitely much faster than VTM as VVEnc took around 1h to encode 6000 frames in UHD, so with a very basic calculation we have 6000 FPH, 100 FPM and around 2 FPS. Taking into account the 1 Gbit/s network slowdown (the source was a 1.5 Gbit/s 12bit 4:4:4 Apple ProRes), the fact that Avisynth indexers like ffms2 are not particularly optimized for it and the pipe of the pipe (Avisynth -> FFMpeg -> VVEnc) that's surprisingly fast for a reference encoder which is not using all the cores.

[Gravitator]

Originally Posted by funky080900

in vvenc bitrate is specified in bits per second rather than kilobits/second. So 25000 is just 25 kbps which is probably not what you wanted.

I got caught too!
--bitrate, --kbitrate, --mbitrate

[MythCreator]

Originally Posted by FranceBB

Where? Your link leads me to an empty folder.
Are you sure you made it public?






Ah, really? That's good to know.



That's a shame, really, as ffmpeg is used as a decoder by many programs and it's the base of our beloved Avisynth indexers and also MPV which is my de facto player of choice.




That's nice, 3fps for FULL HD is acceptable and I think I can use it to finally have a proper test with my UHD Masterfile.
I'll let you know how it goes as soon as I get my hands on a Windows build (if @MythCreator replies)

It's public now, sorry for the inconvenient...

And as to the extremely slow, I think maybe it's something with my script?
vspipe "test.vpy" - | vvencapp.exe -s 1920x1080 --framerate 23.976 -f 500 --threads 16 --preset slower -q 18 -i - --output "test.vvc"

I dithered bit depth to 10 in vpy, the ET for only 1 frame is around 500.

[Gravitator]

Originally Posted by MythCreator

It's public now, sorry for the inconvenient...

Share the decoder?

[MythCreator]

Originally Posted by Gravitator

Share the decoder?

Done.

[FranceBB]

Originally Posted by MythCreator

as to the extremely slow, I think maybe it's something with my script?
vspipe "test.vpy" - | vvencapp.exe -s 1920x1080 --framerate 23.976 -f 500 --threads 16 --preset slower -q 18 -i - --output "test.vvc"

I dithered bit depth to 10 in vpy, the ET for only 1 frame is around 500.

I don't know, it shouldn't make any difference if your pipe comes from VapourSynth and mine from Avisynth, however with me getting around 2 FPS for UHD, I expected you to get more than 3FPS for FULL HD. Still, 2fps is something I can live with as it's dramatically faster than VTM. About the dithering, I have been dithering down the input as well in Avisynth using the Floyd Steinberg error diffusion. Anyway, next week, when I'm gonna have the server free from whatever my coworkers here at Sky are encoding, I'm gonna re-encode a few masterfiles in H.266, H.265, H.264 and MPEG-2 at the same bitrate, namely 25 Mbit/s and see how they behave.

[Gravitator]

Originally Posted by MythCreator

Done.

Thanks!
------
Who has a thought about the reason for the violation of the exit from me?
Enc:

C:\>ffmpeg -i z:/space_6s.mkv -pix_fmt yuv420p -frames 1 -f yuv4mpegpipe - | vvencapp.exe -s 1920x1080 --framerate 23.976 --threads 2 --preset medium -q 30 -i - --output z:\test.vvc

Dec_1:

C:\>vvdecapp.exe -b z:\test.vvc -o z:\exit.yuv

Dec_2

C:\>ffmpeg -video_size 1920x1080 -framerate 23976/1001 -pix_fmt yuv420p10le -i z:\exit.yuv z:\output.jpeg

The problem is solved through the second feeding method:

C:\>ffmpeg -i z:/space_6s.mkv -pix_fmt yuv420p -frames 1 -f rawvideo - | vvencapp.exe -s 1920x1080 --framerate 23.976 --threads 2 --preset medium -q 30 -i - --output z:\test2.vvc

[MythCreator]

Originally Posted by FranceBB

I don't know, it shouldn't make any difference if your pipe comes from VapourSynth and mine from Avisynth, however with me getting around 2 FPS for UHD, I expected you to get more than 3FPS for FULL HD. Still, 2fps is something I can live with as it's dramatically faster than VTM. About the dithering, I have been dithering down the input as well in Avisynth using the Floyd Steinberg error diffusion. Anyway, next week, when I'm gonna have the server free from whatever my coworkers here at Sky are encoding, I'm gonna re-encode a few masterfiles in H.266, H.265, H.264 and MPEG-2 at the same bitrate, namely 25 Mbit/s and see how they behave.

I figured it out lol. --preset slower shouldn't be named as slower, placebo is more accurate. --preset fast is 100 times faster than Slower at least.

[FranceBB]

Originally Posted by MythCreator

--preset slower shouldn't be named as slower, placebo is more accurate.

LOL
It's a good thing that I generally stick with --preset medium then ehehehehehe

[Gravitator]

C:\>ffmpeg -i z:/space_6s.mkv -pix_fmt yuv420p -f rawvideo - | vvencapp.exe -s 1920x1080 --framerate 23.976 --rskip 0 --threads 2 --preset medium -q 30 -i - --output z:\VVC_R0.vvc

I get this:

command line error: Unknown option `rskip' (value:`0')

How to disable --rskip?

[LigH]

Please do not assume that vvencapp supports all the same parameters as x265. You won't find an rskip parameter mentioned in vvencapp --fullhelp

[Gravitator]

Originally Posted by LigH

Please do not assume that vvencapp supports all the same parameters as x265. You won't find an rskip parameter mentioned in vvencapp --fullhelp

Terribly scarce settings
Let's wait for advanced modification of ...

[Funky080900]

There is the “full-featured” version of VVenC which has more settings. Don't know about rskip tho.

[Gravitator]

Originally Posted by Funky080900

There is the “full-featured” version of VVenC which has more settings.

Where is it?

[Funky080900]

You should get both vvencapp and vvencFFapp when following the build instructions.

[LigH]

They still depend on Visual Studio. Or are Jamaika's patches still sufficient for compilation with GCC (maybe in MABS)?

[Gravitator]

A new VVC ecoder at the AV1 level (visually gives 25% better quality than x265).

[birdie]

Originally Posted by Gravitator

A new VVC ecoder at the AV1 level (visually gives 25% better quality than x265).

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. In terms of compression efficiency only AV2 targets VVC but then again at the expense of being more CPU hungry.

I'm curious MulticoreWare has pretty much ceased x265 development (there have been very few commits for the past half a year), yet x266 is nowhere to be seen. Ha! Their x266 GIT repo does exist but it's for authorized users only.

[Gravitator]

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. In terms of compression efficiency only AV2 targets VVC but then again at the expense of being more CPU hungry.

This is a statement on my test.
I expect that disabling --rskip should give VVC: medium a victory over x265: placebo.