Added
- Dispatcher capability gating: QSV and VAAPI jobs only schedule on nodes that advertise the matching encoder capability (no NVENC↔QSV translation — those are vendor-specific hardware paths).
- README and
docs/installation.mdnow have an "Intel/AMD GPU support (experimental)" section with the compose passthrough snippet, verification steps, and troubleshooting for the most commonvainfofailures. Help wanted from anyone who can run it end-to-end.
Docker images
Pinned to this release:
docker pull ghcr.io/i-ial9000/shrinkerr:0.3.70 # portable (amd64 + arm64, CPU)
docker pull ghcr.io/i-ial9000/shrinkerr:0.3.70-nvenc # x86_64 + NVIDIA GPU (NVENC)
docker pull ghcr.io/i-ial9000/shrinkerr:0.3.70-edge # as above but ffmpeg master
docker pull ghcr.io/i-ial9000/shrinkerr:0.3.70-edge-nvencFloating tags (:latest, :edge, :nvenc, :edge-nvenc) track the most recent release of each lineage.