Makes the derived files for every image folder in the VFB image store
(VFB/i/<xxxx>/<yyyy>/<VFB_template>/ and uploads VFBu_<id>/<VFB_template>/):
| File | Made from | Replaces (Jenkins) |
|---|---|---|
thumbnail.png, thumbnailT.png |
volume.nrrd + the template's template.nrrd (NRRDtools create_mip / colorize_image_stack / merge_images, ImageMagick for the alpha) |
CreateThumbnail |
volume.wlz |
volume.nrrd with corner markers ([0,0,0], [max,max,max] ≥ 3) → WlzExtFFConvert -f nrrd → WlzThreshold -v2 -H → WlzSetVoxelSize |
CreateBoundedImageForWLZ → CreateTIFFforWLZ → CreateWLZ |
volume.obj |
link to volume_man.obj when that exists, else a point cloud (derive/pointcloud.py, a vectorised maxProjVol.py — byte-identical without jitter, see tests/) |
CreatePointCloudOBJ |
| — | legacy volume_bounded* intermediates removed |
Clean_bounded_images |
Each volume.nrrd is read from the share once; outputs are written as <name>.tmp-* and renamed into place.
Run on the VFB cluster as the Argo workflow image-derivatives (talosconfig apps/pipeline):
scan (one pod per shard) → plan → derive (batches in parallel, vfb-batch priority) → report.
python -m derive.scan --root data --shard 0010 --out RUN/folders/0010.jsonl.gz
python -m derive.plan --run-dir RUN --state-dir STATE # or --sample 50 for a validation set
python -m derive.worker --run-dir RUN --batch 00000 [--validate] [--clean]
python -m derive.report --run-dir RUN --state-dir STATE
python -m derive.compare --run-dir RUN # after a --validate run