Comment by squeedles

FYI, getting hardware acceleration for Intel working with Jellyfin is pretty straightforward. The key thing is that you have to put the server acct into the render group, then pass through the dri device. I also pass through the video group, but I don't think that is strictly necessary. NVIDIA seems a bit different but I can't speak to that. Docker compose file looks something like the following (uids and gids may vary)

  services:
    jellyfin:
      container_name: jellyfin
      image: jellyfin/jellyfin:latest
      pull_policy: always
      user: "1007:1003"  # jellyfin / jellyfin
      group_add:
        - "44"   # video
        - "992"   # render
      network_mode: bridge
      ports:
        - 8096:8096
        - 8920:8920
      volumes:
        [ ... config, cache, and content ... ] 
      devices:
        - /dev/dri:/dev/dri