Pipeline execution:
tile_compile_runner run \
--config <path> \
--input-dir <path> \
--runs-dir <path>Resume:
tile_compile_runner resume \
--run-dir <path> \
--from-phase <PHASE>Options:
--max-frames <n>— limit frames (0 = no limit)--max-tiles <n>— limit classic tile processing (0 = no limit)--dry-run— validate without full processing--run-id <id>— custom run ID--stdin— read config from stdin
Pipeline phases run from SCAN_INPUT (0) through DONE (18), including
HYPERMETRIC_STRETCH (17). Supported resume entry points depend on available
artifacts; common points are STACKING, ASTROMETRY, BGE, PCC, and
HYPERMETRIC_STRETCH.
GPU selection is configured in YAML, not with a runner flag:
runtime_limits:
acceleration_backend: autoInspect artifacts/acceleration_context.json or the live log fields
cpu_workers, gpu, and backend to confirm the effective backend.
Config:
get-schema— print JSON schemadump-default-config— default config as JSONload-config <path>— load YAMLsave-config <path>— save YAMLvalidate-config— validate
Run inspection:
list-runs <dir>get-run-status <dir>get-run-logs <dir> [--tail N]list-artifacts <dir>
Input:
scan <dir> [--frames-min N]
FITS:
fits-stats <path>
PCC:
pcc-run <in.fits> <out.fits> --wcs <wcs.fits> [--source vizier|siril]pcc-apply <in.fits> <out.fits> [--r X] [--g Y] [--b Z]
GUI state:
load-gui-state [--path <file>]save-gui-state [--path <file>] [--stdin | <JSON>]