[Bug]: QSV/FFmpeg: Filtergraph pixel format negotiation fails on mixed 8bit/10bit inputs (defaults to monow or throws -22 error)

[Astraea35]

Which component impacted?

Video Processing

Is it regression? Good in old configuration?

None

What happened?

Command used:

Bash
ffmpeg -hwaccel qsv -hwaccel_output_format qsv -i input.mp4 -filter_complex "[0:v:0]hwdownload,scale,format=yuv420p10le..." -c:v libsvtav1 output.mkv
Behavior:
When batch processing mixed 8bit (H.264/HEVC) and 10bit (AV1) videos using hwdownload and scale, the FFmpeg QSV static format negotiation layer fails.

For 8bit videos, it weirdly negotiates to monow format, causing hwdownload to crash with Invalid output format monow.

For 10bit videos, it can conflict and throw Invalid output format yuv420p or error code -22.

Unlike NVIDIA's CUDA implementation which handles mixed bit-depths seamlessly, Intel QSV requires rigid manual padding (scale_qsv=format=p010) to avoid filtergraph deadlocks.

3. Environment（环境信息）
   GPU: Intel(R) Iris(R) Xe Graphics

Tool: FFmpeg with QSV/oneVPL enabled

What's the usage scenario when you are seeing the problem?

Transcode for media delivery

What impacted?

It blocks automated batch video transcoding workflows when handling mixed-format media libraries.

Debug Information

No response

Do you want to contribute a patch to fix the issue?

None

[nyanmisaka]

There is no yuv420p10le support in fixed-function hardware. CUDA is GPGPU. They cannot be directly compared.

https://github.com/intel/media-driver/blob/master/docs/media_features.md#supported-video-processing-cscscaling-format