shu4dev's picture
Bowl body from the user's depth profile
da381de verified
|
Raw History Blame Contribute Delete
2.99 kB

hotec_wheatstraw/source

hotec_gen.py is the generator, moved out of the code repo on 2026-10-21 (it was code/src/dishsim/hotec_gen.py) and edited here twice on 2026-10-29: v3 gave the bowl a vertical foot ring; v4 builds the bowl inner-first from the user's measurements (BowlProfile). The inner surface is a PCHIP z(r) through profile_points (the converted depth readings) and the rim edge, mirrored about the axis. The outer surface is that curve offset by the 2.8 mm wall along its normal, then a 0.9 mm straight step to the measured outer rim corner, so the rim top is flat and one wall wide. The foot ring is a plain rib from the foot plane up into the shell; its collision band is buried 1 mm in the shell. The plate and cup profiles are unchanged. Its sha256 260b3aa0bed8b13f4c98f9d2da3f2e78ab5f5174cac2eb15123ecf8451957ed0 is the generator_sha256 stamped in every USD and in catalog.json (v3's file da46d390... and the 2026-10-21 file c7fdeef1... stamped v3 and v2); its docstring still shows the old -m dishsim.hotec_gen form.

Last run (2026-10-29, v4: the bowl from the user's depth profile, rim, wall, height and foot ring, all in the module table, so only the masses are set), on the Mac from the code repo root (code repo plan at 8b2a4fb plus uncommitted edits; the generator lives only in this folder), in a uv environment with Python 3.10, numpy 1.26.0, scipy 1.14.1 and usd-core 24.5, as v3. A second build reproduced every file byte for byte. --no-figures keeps the catalog to the shipped files:

uv run --no-project --python 3.10 --with numpy==1.26.0 --with scipy==1.14.1 --with usd-core==24.5 --with matplotlib==3.9.2 \
    python data/assets/hotec_wheatstraw/source/hotec_gen.py --out <scratch> --no-figures \
    --set plate.mass_kg=0.094625 --set bowl.mass_kg=0.067375 --set cup.mass_kg=0.0595

The same build inside the container: code/util/run_py.sh data/assets/hotec_wheatstraw/source/hotec_gen.py with the same flags. The generator refuses to overwrite, so build into a scratch folder, then copy the three .usda and README.md into ../usd/ and the two JSON files into ../. Then regenerate the dish manifests with python3 code/src/sampler/writers/hotec_wheatstraw.py (code repo). Tests:

code/util/run_py.sh -m pytest data/assets/hotec_wheatstraw/source -q      # test_hotec_gen.py, Kit-free

On the Mac the same pytest runs in the uv environment above plus --with pytest (19 passed, 1 skipped: the USD round trip needs the container).

v3 (2026-10-29) was the v4 command plus --set bowl.foot_diameter_m=0.06572 --set bowl.foot_width_m=0.00316 --set bowl.foot_height_m=0.00557 --set bowl.height_m=0.07312, run with the generator da46d390.... v2 (2026-09-22) was code/util/run_py.sh -m dishsim.hotec_gen --out data/assets/models/hotec_wheatstraw/v2 --set plate.mass_kg=0.094625 --set bowl.mass_kg=0.067375 --set cup.mass_kg=0.0595 (corallab container; the generator file committed at 9e3b163).