Beiträge von Vouk

[12:52:27] Options:         _pixelFormat=p010le multipass=disabled preset=p7 qp=20 rc=constqp

Are you sure you want to encode in 10 bit and not 8 bit?

It'd help to use a faster preset. p2 and p1 would even use both encoding engines of the 4090.

For some reason the cpu is the bottle neck here.

[12:53:01] Frame #171: vRender: 695781 us, vProcess: 15423 us, vEncoding: 21569 us, aRender: 116 us, aEncoding: 1115 us, Latency: 735244 us
[12:53:01] Frame #172: vRender: 21765 us, vProcess: 10612 us, vEncoding: 18021 us, aRender: 84 us, aEncoding: 2170 us, Latency: 54566 us
[12:53:01] Frame #173: vRender: 98124 us, vProcess: 18532 us, vEncoding: 25001 us, aRender: 171 us, aEncoding: 14 us, Latency: 143620 us
[12:53:01] Frame #174: vRender: 363 us, vProcess: 13767 us, vEncoding: 21068 us, aRender: 92 us, aEncoding: 1331 us, Latency: 39040 us
[12:53:01] Frame #175: vRender: 3343 us, vProcess: 10310 us, vEncoding: 18882 us, aRender: 192 us, aEncoding: 565 us, Latency: 36057 us
[12:53:01] Frame #176: vRender: 65775 us, vProcess: 15046 us, vEncoding: 30791 us, aRender: 137 us, aEncoding: 2444 us, Latency: 116458 us
[12:53:02] Frame #177: vRender: 837611 us, vProcess: 11949 us, vEncoding: 23436 us, aRender: 244 us, aEncoding: 1192 us, Latency: 876397 us
[12:53:02] Frame #178: vRender: 218 us, vProcess: 9556 us, vEncoding: 17262 us, aRender: 86 us, aEncoding: 6 us, Latency: 28865 us
[12:53:02] Frame #179: vRender: 37892 us, vProcess: 10338 us, vEncoding: 15881 us, aRender: 89 us, aEncoding: 575 us, Latency: 66451 us
[12:53:02] Frame #180: vRender: 47363 us, vProcess: 18529 us, vEncoding: 34612 us, aRender: 124 us, aEncoding: 1708 us, Latency: 104267 us

vRender (The time premiere needs to render the frame) is constantly changing between 0.2 ms and almost 700 ms. What's your cpu doing?

Okay, I improved it even more. You can now even execute JavaScript in the file name using

${...}

So you'll have all the freedom you want for dynamizing your filename.

Improved this a bit more.

Great! Nice to hear!

Found it. For the hwupload_cuda filter the parameter name is device not gpu. Fixed it in 0.7.5.

You might have to edit each hwupload_cuda node, select the gpu and save the scene again to make it work.

Yes, as far as I know there are also USB3 to PCIe x1 adaptors (for mining equipment).

Can you test such a complicated workflow that includes 2 or 3 gpus with plain ffmpeg.exe? Just to check if it is actually an FFmpeg limitation?

Zitat von Joelson_Forte

This issue occurs because there is no preset saved in Scene Design.

This is fixed in 0.7.4.

I'll investigate this issue.

Zitat von buddhabas

Anyway, when I click this botton nothing happen. Why?

Difficult to tell. Normally the Scene Designer should open. Does it open if you start it from the desktop icon?

Yes, I fixed this in the next version so $(OutputFileName) does not contain the extension anymore.

Edit: This is fixed in 0.7.4.

I'm absolutely not familiar with this render server setup.

Can you give me instruction to setup a system like this so i can develop for it and test on it?

Added.

Zitat von Joe24

So there is no actual problem.

Nice!! Well, you've still got the improvement of a named device dropdown instead ofentering the device number

Zitat von Joe24

It would certainly be nice to run multiple GPUs from a single Vegas render, but if this is in fact an FFmpeg limitation, I guess there's not much anybody can do about it except to run multiple Vegas instances, each controlling it's own GPU. Unless VoPro could run multiple FFmpeg instances from the same Vegas output buffer?

Voukoder(Pro) doesn't use an FFmpeg binary (as most other tools do), it's using its DLL variants. So for each export it creates a new instance. But you can test your filter chain witz the command line ffmpeg version.

In line 65 of your previousely posted log file you can see the text representation of the filter chain:

buffer@9309cf357e4c40e3b1334754fe21a3f3=width=1920:height=1080:pix_fmt=yuv420p:time_base=1001/24000:pixel_aspect=1/1,hwupload_cuda[673157b8b5c94fcbb6460873239f5e54];[673157b8b5c94fcbb6460873239f5e54]split=2[split_74972bd9eab84782948d76fe847c9664][split_21f42936466e4121b3817c10e2155724];[split_74972bd9eab84782948d76fe847c9664]scale_cuda=w=1920:h=1080:interp_algo=3:force_original_aspect_ratio=1,hwdownload,hwupload_cuda=gpu=1,format=pix_fmts=cuda,buffersink@a5bebec201e640c6940a6f7d9caca738;[split_21f42936466e4121b3817c10e2155724]scale_cuda=w=1280:h=720:interp_algo=3:force_original_aspect_ratio=1,format=pix_fmts=cuda,buffersink@bc23b883870a452fa61cea7d0d361e4c

Just append it to the ffmpeg.exe command line using -filter_complex.

Zitat von Joe24

it would make more sense to have each node 'inherit' the GPU assignation of the previous (upstream) node

Yes, and setting the encoder pixel format automatically to CUDA. I might add that later.

Please check out 0.7.3. Please tell me if this works. (I don't have multiple gpus in my machine)

Please check out version 0.7.3.

Supporting these variables in the file name:

Please check out version 0.7.3.

I have already implemented something like this which you will be able to test soon.

Just want to finish the other tasks and then do a new build.

Encoders:

Upload filter:

I'll look into this again.

Thanks for pointing out these edge cases. It's really helping to improve the software.

Will look into that.

Yes, I will add VP9.