Video Transcoding

Note: This is contributed content intended for advanced users. You can contribute by clicking to send a pull request with your changes.

The encoder used by FFmpeg can be configured within your docker-compose.yml config file.

Experimental hardware accelerated transcoding on a Raspberry Pi (and compatible devices) may be enabled using the h264_v4l2m2m encoder:

PHOTOPRISM_FFMPEG_ENCODER: "h264_v4l2m2m"

It defaults to libx264 if no value is set or transcoding with h264_v4l2m2m fails.

Please refer to the official documentation for a full list of encoders and their implementation status.

Info

Some server configurations, specifically Raspberry Pi's, may run into memory allocation issues when using hardware acceleration. Monitor your server logs carefully and increase available GPU and/or CMA memory allocations if necessary.

The Docker container will also need access to one or more video devices. For the h264_v4l2m2m encoder on a Raspberry Pi, also add:

devices:
 - "/dev/video11:/dev/video11"

Additional advanced configuration options:

PHOTOPRISM_FFMPEG_BUFFERS: "64" # FFmpeg capture buffers (default: 32)

Warning

Hardware accelerated transcoding within Photoprism is currently experimental. It may not work on all server configurations. Some users report unexpected hangs in the FFmpeg process when attempting to transcode large video files.