feat(showcase): add grain-sacks

[TMHSDigital]

Adds showcase/grain-sacks, a pile of three burlap grain sacks. Two stand and one lies slumped in front. Each is gathered at the neck under a two-turn twine tie with a pleated crown, settled onto a flat base, and pressed into its neighbour.

Selection: it is the first unclaimed candidate in the ROADMAP pool now that rope-bridge has shipped. It is a prop, so it goes in showcase. There were no open PRs; the only feat/showcase-* refs are already-merged work.

Binaries (all local, .scratch/):

• blender-4.5.11-windows-x64, reports Blender 4.5.11 LTS
• blender-5.1.2-windows-x64, reports Blender 5.1.2
• blender-5.2.1-windows-x64, reports Blender 5.2.1 LTS

The budget that matters: contact patch (exit 19)

A sack full of grain settles onto a flat patch. An egg-bottomed sack still touches the floor at one vertex, so both zmin and the per-sack support gate pass it. The check sums, per sack, the area of faces lying flat on the floor.

• Default: 0.035 / 0.099 / 0.117 m², against a 0.012 floor.
• --round-bottom: keeps the same sink and fold height and every sack's single lowest point on the floor. It exits 19 with every patch at 0 while zmin and the supports pass. live-run-proven

Falsifiers: exit codes per binary (live-run-proven)

	default	--skip-decimate	--stray-vert	--lift-z	--float-sack	--slip-tie	--loose-tie	--part-sacks	--round-bottom	--high-belly	--output
4.5.11	0	9	15	16	16	17	18	18	19	19	—
5.1.2	0	9	15	16	16	17	18	18	19	19	—
5.2.1	0	9	15	16	16	17	18	18	19	19	0

Flag	Target budget	Measured	Exit
--skip-decimate	LOD1 ratio	1.0000	9
--stray-vert	hygiene	loose_v 1	15
--lift-z	grounded zmin	0.050	16
--float-sack	named supports	worst 0.012	16
--slip-tie	tie in the neck waist	margin −13.4 to −44.6 mm (need ≥ +8)	17
--loose-tie	tie seat band	−1.67 mm	18
--part-sacks	press between sacks	−4.0 mm	18
--round-bottom	contact patch	0 m² × 3	19
--high-belly	settled belly	0.535–0.586 of the body (max 0.42)	19

Envelope: every falsifier stays inside the 20 mm BBOX_TOL. The worst cases are --high-belly at −16.2 mm in Y and --part-sacks at +15.5 mm in Y. live-run-proven

Checks run locally

• tests/check_falsifier_targets.py: static check passes. --run <BLENDER> --only grain-sacks passes all 9 on each of the three binaries. live-run-proven

• tests/smoke/run_example.py --forbid-skip: [PASS] grain-sacks on 4.5.11, 5.1.2 and 5.2.1. live-run-proven

• Other local checks:

  • check_exit_code_readme, check_import_export_rules, check_engine_id, check_gallery_images, test_harness
  • the inline validate-manifest and validate-counts jobs, run in a clean worktree (30 showcase pieces)

  All pass. live-run-proven

• Determinism: every measured line is byte-identical across 4.5.11, 5.1.2 and 5.2.1, including both LOD counts and the glTF byte count. The piece uses no RNG at all. Check-only wall-clock is about 1.7–2.1 s. live-run-proven

• Ecosystem drift check is a remote action, so it is CI-proven only. The 4.5 / 5.1 / 5.2 smoke jobs are CI-proven on this PR.

Findings the budgets and inspection forced

• Coplanar pairs. The coplanar budget counted 2 cross-shell pairs: a twine face lay parallel to the cloth face it hugs, one column over. The tie turns now sit a quarter step off the body's columns, and the count is 0.
• Waist measure. The first waist measure (narrowest axial bin) read 32.6 mm off on the lying sack, whose rings straddle bins. It was replaced by a local-minimum test: the host under the tie must be ≥ 8 mm narrower than 40 mm either side. The probe band was widened to ±12 mm after an empty band read as radius 0.
• Collider. The per-sack hull sampled in shell traversal order (1336 tris). It now strides rings and columns in build order and comes to 376 tris.
• Draft render, round 1:
  • The 6 mm weave aliased into moiré, so the period is now 12 mm with a softer bump.
  • The stripes sat on the standing sacks' hidden face, so each sack now has a front and the cloth coordinate is centred on it. inspection-only

Visual gates

Contact sheet: docs/gallery/contact-sheets/grain-sacks-contact-sheet.webp

criterion	grain-sacks	calibration	verdict
mean luminance	0.2513	0.151–0.302	in band
stage darkness	0.2282	0.145–0.236	in band
wedge warmth	+0.185	−0.43 … +0.37	warm
saturation	0.123	0.144–0.231	just under the set; in line with the timber/cloth showcase peers (0.09–0.136)
subject fill	0.789 tall	—	passes the framing gate, min margin 0.072
thumbnail legibility	—	—	three tied sacks read at tile size

Asset sheet: docs/gallery/asset-sheets/grain-sacks.webp. Verdict: passes. Rendered alone under neutral light, the pile carries more than the rocket or the arm: weave, pleated crowns, twine ties and stripes. It sits level with the hydrant. The helper's standard camera clips the lying sack's base at the panel edge; that is the helper's framing rule, not the model. inspection-only

Close-ups (neck ties, sack-to-sack press, ground contact, back, top, and a clay pass of each): the ties hug the pleated necks, the bases sit flat, and the press reads as a clean seam. There is no z-fight speckle and no faceting on the bodies. The crown tip is slightly pointed, which I judged acceptable. inspection-only

The hero was rendered on 5.2.1 through scripts/render_hero.py, and build_gallery.py was run last. I read the generated alt text and witness callout in docs/gallery/index.html and docs/gallery/grain-sacks/index.html. inspection-only

🤖 Generated with Claude Code