Commit e00d72a
fix(showcase): remodel fern-mossy-rock with a cleaved gritstone boulder and feathered, lobed moss
The boulder read as a smooth grey lump and the moss as a carpet with a hard
vertical front edge; the piece sat wide and low in its panel. Rebuilt:
- boulder: ellipsoid 0.74 x 0.60 x 0.80 (was 0.56 high), cut by twelve deep
planes (0.50-0.58 of the support, was 0.64-0.72) through 16-20 mm bevels
(were 35-110 mm), so broad flat fracture faces meet at crisp, weathered
arrises; two stepped ledges (region beyond both of two planes removed);
flat faces keep a tenth of the waves (was 40%); lattice 28 -> 36 (high
40 -> 52). Crack plane moved to the stone's middle so the V reads down its
front, with the crack fern growing out of it in view of the hero camera
- stone shader: warm buff to grey mottling, weathered grey rind, quartz grit,
pitting in the bump, rain streaks on steep faces only, algae, sage and
orange lichen and dark crust spots, a green stain round every cushion
(MossD point attribute, distance to nearest moss) and a dark damp band at
the foot (SoilH, height over the bank)
- moss: lobed outline (broad lobes, deep inlets); height full in the middle,
feathering to nothing over the outer 42% of the radius, up to 1.8x deeper
in the stone's hollows; two bare patches dip under the stone; the rim runs
over the crack's lip; nine satellite tufts; each shell tucked to its own
depth (two rims on one flat fracture face were coplanar)
- composition: crown moved to the boulder's front-right foot, soil disc
0.95 x 0.78 (was 1.12 x 1.03), camera 4.0 -> 3.55 m; litter 48, pebbles
14, sorrel 7 clumps, placed sorrel first so the tighter disc holds them
Budget changes:
- Base tris 76800-77600 -> 86300-87200 (86740); face floors re-fit (16224
stone, 4548 moss, 1728 litter, 2150 sorrel measured)
- Outer AABB -> (2.2872, 1.9008, 1.1111) m; the crack fern is the top, the
soil disc the left and back, frond tips the right and front
- Moss seat band 0.006-0.080 -> 0.001-0.055 (0.0016-0.0463); bare-patch
vertices are tagged Ring -2 and measured inside the stone with the rim
- New exit 27, cleaved: share of the stone's turning (dihedral x edge
length, >20 mm over the soil, crack excluded) in arrises sharper than 20
deg >= 0.42 (0.4856). A count of flat 0.03 m2 regions did not separate
the recipes (the old lump had as many); where the turning happens did
- New exit 28, moss rim: each cushion's last ring before its tucked rim
<= 0.010 m over the stone (0.0069)
Falsifiers (flag -> measured -> exit), 5.2.1 and 4.5.11 identical; tris
86740 and AABB unchanged in all:
--skip-decimate LOD1 1.0000 9
--stray-vert loose_v 1 15
--lift-z zmin 0.05000 16
--smooth-rock crisp share 0.2892 (new; shipped recipe,
ellipsoid x0.90, own flat foot) 27
--float-fronds stalk base -0.0503 m 17
--lift-crown rootstock worst sector -0.0180, 1/8 17
--float-moss top 0.0858, rim -0.0380 m 18
--slab-moss rim ring 0.0189 m (new; shipped profile) 28
--perch-rock worst sector -0.0321 m, 7/8 20
--sunny-moss facing 0.5986 21
--perch-crack-fern not in cleft, walls 0.1576 m 22
--flat-taper tip 0.996 23
--opposite-pinnae gaps 0.0013-0.9987 23
--open-crozier 0.932 turns, ratio 1.128 24
--float-cover -0.046 m, 164 shells loose 25
tests/check_falsifier_targets.py --run --only fern-mossy-rock passes.
Evidence (live runs, local):
- .scratch/5.2/blender-5.2.1-windows-x64/blender.exe (Blender 5.2.1 LTS):
two default runs identical; hygiene 0 incl. 0 coplanar cross-shell pairs,
0 pinnae unturned; collider 130; export 10731436 B.
- .scratch/blender-4.5.11-windows-x64 (4.5.11 LTS): all 17 runs same codes
and values; glTF +12 B.
- .scratch/blender-5.1.2-windows-x64 (5.1.2): default exit 0, same values.
- Framing: fill x 0.822, y 0.878; margins l .131 r .047 b .044 t .078.
- Contact sheet: luma 0.2387 (band 0.2369-0.2428), stage 0.2428 (<= 0.2517),
warmth +0.0949 (spread -0.144..+0.105), sat 0.2371. Thumbnail: split
mossy boulder with a fern in its crack beside the fern crown.
- Asset sheet: before, a grey lump under a moss slab behind a fern; now a
split, blocky boulder with lobed moss and the crack fern. Not the
least-designed object (the hydrant is plainer). Weak points: bare moss
patches read a little like cut holes up close; the soil rim is a thick
cake edge; the rock stays in shade in the hero.
- build_gallery, check_site_links (122 pages, 0 failures), check_engine_id,
check_exit_code_readme, check_gallery_images, check_import_export_rules,
check_falsifier_targets pass.
Co-Authored-By: Claude Opus 5.5 (1M context) <noreply@anthropic.com>1 parent b1c9b78 commit e00d72a
11 files changed
Lines changed: 1077 additions & 345 deletions
File tree
- docs/gallery
- asset-sheets
- assets
- contact-sheets
- fern-mossy-rock
- showcase
- fern-mossy-rock
| Original file line number | Diff line number | Diff line change | |
|---|---|---|---|
| |||
174 | 174 | | |
175 | 175 | | |
176 | 176 | | |
177 | | - | |
| 177 | + | |
178 | 178 | | |
179 | 179 | | |
180 | 180 | | |
| |||
304 | 304 | | |
305 | 305 | | |
306 | 306 | | |
307 | | - | |
| 307 | + | |
308 | 308 | | |
309 | 309 | | |
310 | 310 | | |
| |||
0 commit comments