Interlaced video transcoding issues

[LukaszBrzyszkiewicz]

Hi,

when I'm trying to transcode interlaced video I have issues different vs how I'm trying to do workaround. Still I didn't figured out working solution.

1. When I'm using GPU based deinterlacer I have cutted in half audio QSVEncC64 --codec hevc --avhw --quality best --output-depth 10 --gop-ref-dist 16 --vpp-pmd apply_count=1 --vpp-detail-enhance 50 --vpp-edgelevel strength=1.0 --avbr 2000 --chapter-copy --video-metadata clear --metadata clear --audio-codec 1?libopus:mapping_family=1 --audio-stream 1?5.1 --audio-bitrate 1?256 --metadata language=hu --device 2 --vpp-deinterlace bob

after receiving error in the middle of transcoding of a file: cop.SingleSeiNalUnit value changed off -> auto by driver QSVEncC (x64) 7.93 (r3764) by rigaya, Jul 16 2025 12:20:51 (VC 1944/Win) OS Windows 11 x64 (26100) [UTF-8] CPU Info 13th Gen Intel Core i5-13600K [5.12GHz] (6P+8E,14C/20T) <DG2> GPU Info Intel Arc A380 Graphics (128EU) 300-2450MHz (32.0.101.6913) Media SDK QuickSyncVideo API v2.15, FF, 2nd(d) GPU Async Depth 3 frames Hyper Mode off Buffer Memory d3d11, 96 work buffer Input Info avqsv: H.264/AVC, 1920x1080, 25/1 fps VPP ColorFmtConvertion: nv12 -> p010 Deinterlace (bob) cspconv(p010 -> yv12(16bit)) denoise(pmd): strength 100, threshold 100, apply 1, exp 1 edgelevel: strength 1.0, threshold 20.0, black 0.0, white 0.0 cspconv(yv12(16bit) -> p010) Detail Enhancer, strength 50 AVSync auto Output H.265/HEVC(yuv420 10bit) main10 @ Level 4.1 (high tier) 1920x1080p 1:1 50.000fps (50/1fps) avwriter: hevc, #1:ac3/5.1(side):5.1 -> libopus/5.1/256kbps => matroska Target usage 1 - best Encode Mode Quality VBR bitrate Bitrate 2000 kbps Max Bitrate 15000 kbps QVBR Quality 23 VBV Bufsize 30000 kb QP Limit min: 22, max: 63 Ref frames 5 frames Bframes 15 frames, B-pyramid: on Max GOP Length 500 frames Ext. Features WeightP WeightB GPB ctu:64 sao:all avout: avcodec writer: failed to send frame to audio encoder #1: Invalid argument out size 703.7MBB avout: avcodec writer: failed to send frame to audio encoder #1: Invalid argument avout: avcodec writer: failed to send frame to audio encoder #1: Invalid argument

2. Trying --vpp-yadif mode=bob gives video truncated in half :)
3. Trying --vpp-yadif without selecting mode, I have audio truncated in half

So for now probably only WA is to transcode video, and then transcode audio outside QSV and merge together later, but it looks like some FPS calculations are not matching video and audio.

[rigaya]

I also assume that the timestamp calculation within the files is getting messed up, especially for H.264 interlaced input, but the problem is that the sample H.264 interlaced files I have, I get no problem.

When you have half audio in 1,

1. Is the video converted to the end?
2. What is shown for duration and framerate(for video) in MediaInfo for both video and audio in input file?
3. What is shown for duration and framerate(for video) in MediaInfo for both video and audio for output file?

Anyway it seems to depend on the input file, so might be difficult to debug without the actual file.