I-IAL9000/shrinkerr v0.5.26

on GitHub

Fixed

• Conversion failing on 10-bit H.264 sources (Hi10p, yuv420p10le) with ffmpeg exited with code 218 and "Impossible to convert between the formats supported by the filter 'graph -1 input from stream 0:0' and the filter 'auto_scale_0'". Pascal NVDEC (and older) don't support 10-bit H.264; QSV decode is 8-bit-only for H.264; most VAAPI drivers don't either. Pre-v0.5.26 our codec gate only checked the codec name (h264 ∈ supported set → True), so the cmd builder emitted -hwaccel cuda -hwaccel_output_format cuda and a scale_cuda=format=p010le filter. ffmpeg then tried NVDEC, silently fell back to software decode (good), but the now-CPU frames couldn't bridge to the still-present CUDA filter (bad). Extended hw_decode_supports() to also check source_pix_fmt: any H.264 source with a p10 / 10le / p12 / 12le pix_fmt now takes the pure-software-decode path on all three HW backends. 10-bit HEVC is unaffected (Pascal+ NVDEC, Gen11+ QSV, and most VAAPI drivers handle it fine). Worker log now also reports the source pix_fmt alongside the codec when HW decode is skipped.

Docker images

Pinned to this release:

docker pull ghcr.io/i-ial9000/shrinkerr:0.5.26         # portable (amd64 + arm64, CPU)
docker pull ghcr.io/i-ial9000/shrinkerr:0.5.26-nvenc   # x86_64 + NVIDIA GPU (NVENC)
docker pull ghcr.io/i-ial9000/shrinkerr:0.5.26-edge    # as above but ffmpeg master
docker pull ghcr.io/i-ial9000/shrinkerr:0.5.26-edge-nvenc

Floating tags (:latest, :edge, :nvenc, :edge-nvenc) track the most recent release of each lineage.