First off I'd like to thank Selur - I've been using Hybrid for years without any problems.
I'm on version 2024.12.21.
I've recently begun experimenting with Topaz Video AI to enhance old dvds. The workflow I settled on is 1) deinterlace in Hybrid, 2) process in Topaz, and 3) re-encode in Hybrid. FFV1 420 10 bit for steps 1 & 2 works fine, but when I try ProRes 422 HQ, the final re-encode in Hybrid instantly crashes with the call "Helper1, exitcode: -22". I suspect it's a color matrix issue. All the affected files list the input and output matrix as "?" in the Filtering tab.
For the moment I've stopped exporting as ProRes in Topaz, but a number of dvds are already done, & reprocessing as FFV1 will take many hours. So if that could be avoided I'd be very grateful.
Code:
"C:\Program Files\Hybrid\64bit\ffmpeg.exe" -y -loglevel fatal -noautorotate -nostdin -threads 8 -ignore_editlist true -i "C:\Users\singh\Videos\tas5.mov" -map 0:0 -an -sn -vf scale=out_color_matrix= -pix_fmt yuv420p10le -strict -1 -fps_mode passthrough -f yuv4mpegpipe -
does miss a output_color_matrix value.
Since I can't find 'output_color_matrix' in the source any more: Does this also happen if you try the same with the latest dev?
Cu Selur
I'm sorry I don't see any version later than rev 2024.12.21.1 on the downloads page, so I think I'm already on the latest dev?
Mediainfo list the matrix coefficients of the source file as "identity".
You are not.
↓↓↓↓↓↓↓↓↓↓↓↓↓↓↓↓↓↓↓↓↓↓↓↓↓↓↓↓↓↓↓
Same result in 2025.02.11.[attachment=2994]
Wrong DebugOutput?
The debug output you send was created with:
Code:
2025.02.12 - 15:50:10_Windows 11 Version 24H2 (64bit)_2024.12.21.1 - level 9
Cu Selur
Ps.: In case you can really reproduce the problem with the latest dev, please also share a short sample of the file. (check that the sample also triggers the problem)
Sorry, I must have mixed up the zip files in the attachments. The error still occurs.
Here's a 7 second
sample of a different dvd with the same problem; IME the error occurs with all Topaz exports in ProRes.
I'll look at it.
I can reproduce the problem.
Cu Selur
Not totally sure how I should handle it properly atm.
Source says: ColorMatrix is "identity", but Svt complains that: "Identity matrix (matrix_coefficient = 0) may be used only with 4:4:4 color format."
which isn't supported in main profile,..
Quote:Commercial name : ProRes
Format version : Version 0
Format profile : 422 HQ
....
Color space : RGB
Chroma subsampling : 4:2:2
WTF? Libav decodes is as RGB, so it's not 422 HQ, or it's a but in libav and the source isn't RGB.
-> that makes no sense. Seems like Topaz is creating invalid files,..
Cu Selur
ffmpeg reports it as:
Code:
Stream #0:0[0x1]: Video: prores (HQ) (apch / 0x68637061), yuv422p10le(gbr/bt470bg/reserved, progressive), 1440x812, 324383 kb/s, SAR 1:1 DAR 360:203, 59.94 fps, 59.94 tbr, 19001 tbn (default)
"(gbr/bt470bg/reserved, progressive)"
gbr → This suggests an incorrect or unknown color primaries interpretation. It typically means FFmpeg is reading the format metadata incorrectly or the input file has an unusual flag
bt470bg → This refers to the ITU-R BT.470 color primaries (used for PAL/SECAM).
reserved → This usually indicates an undefined transfer function, meaning the file doesn’t explicitly define how brightness levels are mapped.
progressive → The video is progressive scan (not interlaced).
=> those files are badly flagged
Code:
ffmpeg -y -noautorotate -nostdin -threads 8 -ignore_editlist true -i "C:\Users\Selur\Desktop\sample qtgmc_1_prob4.mov" -map 0:0 -an -sn -pix_fmt yuv420p10le -strict -1 -fps_mode passthrough -f yuv4mpegpipe NUL
results in:
Code:
[swscaler @ 00000234f6d59f80] Unsupported input (Error number -129 occurred): fmt:yuv422p10le csp:gbr prim:bt470bg trc:reserved -> fmt:yuv420p10le csp:gbr prim:bt470bg trc:reserved
[vf#0:0 @ 00000234f530cf80] Error while filtering: Error number -129 occurred
[vf#0:0 @ 00000234f530cf80] Task finished with error code: -129 (Error number -129 occurred)
[vf#0:0 @ 00000234f530cf80] Terminating thread with return code -129 (Error number -129 occurred)
argh,..