Hi everyone, I need help resetting my transcoding state. Everything was working fine with software transcoding, but after a failed attempt to enable Hardware Acceleration, the process is now broken even after reverting to original settings.
System Specs:
Hardware: Mini PC Ryzen 7 5825U (Radeon iGPU), 16GB RAM.
OS: Ubuntu 24.04 (Docker).
Source Files: Honor Magic 7 Pro (Android 15), 1080p 30fps, HEVC "Smart Encoding" enabled.
The Issue:
I enabled VA-API acceleration in the UI (tried both "Auto" and "VA-API" manual settings).
I updated my docker-compose.yml following the official hwaccel.transcoding.yml, mapping the device as - /dev/dri:/dev/dri for both immich-server and immich-microservices.
Transcoding consistently fails with Exit Code 234 and Invalid color space errors. It seems FFmpeg struggles with the Honor's metadata: pixfmt:yuvj420p ... csp:reserved range:pc.
Current Status:
I reverted everything to Software Transcoding (H.264, 1440p, Hable Tone Mapping), restarted containers, and cleared the jobs. However:
New jobs still fail with the same Invalid color space error.
The UI shows ENOENT errors because it tries to access the .mp4 files in encoded-video that failed to generate.
Logs:
[graph -1 input from stream 0:1] Invalid color space
pixfmt:yuvj420p tb:1/90000 fr:30/1 sar:0/1 csp:reserved range:pc
Error reinitializing filters! Task finished with error code: -22
Unable to send file: Error: ENOENT: no such file or directory
I am attaching my docker-compose.yml and the logs below. Has anyone dealt with this specific csp:reserved issue on newer Android 15 devices?