Selur's Little Message Board
Esxi running macos For selur - Printable Version

+- Selur's Little Message Board (https://forum.selur.net)
+-- Forum: Talk, Talk, Talk (https://forum.selur.net/forum-5.html)
+--- Forum: Small Talk (https://forum.selur.net/forum-7.html)
+--- Thread: Esxi running macos For selur (/thread-1495.html)



RE: Esxi running macos For selur - Selur - 29.03.2021

About the length: need a debug output to tell more, wild guess would be that it's vfr related,..
about the angle: seems like a value reset. -> will look into it

Cu Selur


RE: Esxi running macos For selur - shijan - 29.03.2021

Seems it is a bug in last version of ffmpeg-101748-g797c2ecc8f i use. I install ffmpeg 4.3.2 and there is no problem with incorrect file length.

Setting CFR in Config -> input didn't help.


RE: Esxi running macos For selur - Selur - 29.03.2021

Quote:Seems it is a bug in last version of ffmpeg-101748-g797c2ecc8f i use.
Does the version Hybrid is currently packaged with work fine?
Do you see any errors when calling the encoder call in a terminal?

Quote:Setting CFR in Config -> input didn't help.
That would only help if the source is vfr.

Cu Selur


RE: Esxi running macos For selur - shijan - 29.03.2021

ffmpeg version 2020-12-11, originally placed in Hybrid dev version also generates incorrect file length in ProRes.
No problem only with ffmpeg 4.3.2 release version.

Also i see another startup errors:

Couldn't set property(2): value in cropHeight - FilteringModel to 0, found: 48
Couldn't set property(2): value in cropWidth - FilteringModel to 0, found: 48



RE: Esxi running macos For selur - Selur - 29.03.2021

Anything special about your ProRes input file?

Cu Selur


RE: Esxi running macos For selur - shijan - 29.03.2021

I use same ProRes patterns as i share in this folder https://www.dropbox.com/sh/bak63hnr7mnpgbj/AAAf-rMK0LvHYTAFjPuQaaxFa?dl=0

Maybe problem could be due too small file length? They are about 0.6 seconds length.
Also checked - no problem with ffmpeg 4.3.1 and 4.3.2 release version. Bug exists only in ffmpeg development version.

If it is not a problem on Hybrid side, i can start bug report ticket at FFmpeg tracker.


RE: Esxi running macos For selur - Selur - 29.03.2021

Tried with 'Ramp Resolve ProRes 422 HQ 10-bit' on Windows.
Input and output frame count and play links match. (both when using ProRes and H.264 as video format)
-> I'll try to reproduce the issue on MacOs.

Cu Selur


RE: Esxi running macos For selur - shijan - 29.03.2021

Cineform HD also gives incorrect length. Note that MediaInfo reports that frame count is the same, but file length duration is increased.

[Image: iEOcGOf.jpg]


RE: Esxi running macos For selur - Selur - 29.03.2021

That does not seem like a bug in ffmpeg.

If there are 15 frames @25 fps (=1 frame plays 40ms) the correct play length is 600ms.
Only way to get more would be if the source is vfr or not properly flagged.

Ahhh,... I see the problem.
The last frame has a duration of 24s 960ms which is the missing difference. (so the source is vfr)
In case this is properly recorded in the time codes, enabling "Config->Automation->On Load->Always extract timecodes from intput" (and disabling the 'only some') might help.

This effect might also happen if there is an addition stream that is 25s 560ms long.
Iirc there was a change in ffmpeg that it properly stops decoding if there are no additional frames. Older versions returned the last frame over and over again until the length was archived.

-> this is not a bug in Hybrid and personally I wouldn't even call it a but in ffmpeg.
https://superuser.com/questions/1439924/ffmpeg-creates-an-ouput-file-shorter-than-its-longest-stream sounds similar to your issue.

Cu Selur


RE: Esxi running macos For selur - shijan - 29.03.2021

Also with ffmpeg 4.3.2 i also can't render to CineformHD. But this time it is a large mp4 test file.
"ERROR: /Hybrid Temp/44.1hz test source_2021-03-29@19_44_31_0210_01.mov was not created!"

No problem with ffmpeg development snaposhot version
Here is debug