Review-only meadow material candidate

Publication update: Root inspected Comparison.png and explicitly approved publication. The reviewed mask 8f214436…24fe is now live, and the canonical summer-mask generator reproduces it exactly through approved-mask.json. Native image acceptance remains pending. The preparation-stage text below records the original review-only scope.

The actual 19:54 Alps01 view0 was inspected alongside the original verified Grass001 and Grass005 colour files. The native valley is still too uniform and lime-green. Grass005 itself is substantially brighter than Grass001; the current valley blend is narrowly gated, leaving large eligible meadow areas almost entirely on Grass005. No source colour correction is needed to test a darker, more varied meadow blend.

No live Assets files were changed. The candidate and its review comparison are under ArtSource/Alps/SummerMeadow/MeadowCandidate/:

• Comparison.png: original source textures, a 65% Grass001 linear-light source blend, and before/after top-down material albedo. This is a calculated unlit material preview with average source texture mip values, not a game frame or an edited screenshot.
• SchneebergLayers_MeadowCandidate.png: candidate mask, SHA-256 8f214436fbcf4fb68bab779126387d662b229c4fc4cd16ed5968b501288624fe.
• SchneebergLayers_Previous.png: exact frozen reviewed input 623b4d9ec6aa012f7af1b3bbd253404c599db951c7a0c9cd672b58272dc5b85e.
• MeadowEligibility.png, manifest.json and prepare.py: eligibility, source/output hashes, numeric checks and reproducible generation. Reruns use the frozen input, so publication cannot compound the blend.

The change affects R only. Eligibility excludes woodland (original probability ≥0.38), substantial geological/canopy coverage and steep or high land. It fades across original forest probability 0.16–0.38, geology 0.08–0.30, canopy 0.05–0.25, slope normals 0.80–0.96 and local altitude 650–900m. This is aimed at low, gentle non-forest meadows, including the Alpine valley; it does not recolour the limestone massif, forest interior or winter.

Within eligible ground, the two unchanged materials receive coherent variation at several scales: elongated, gently curving meadow parcels with transitions around 200–450m, a restrained darker contribution in smoothed terrain depressions, and subtle 50–90m variations resolved by the approximately 6m mask texels. There are no scattered random spots, a global multiplier, oversized grass texture or new layer. Both original materials, normals and roughness retain their existing 1.4m scale. These are authored agricultural/ground variations, not a claim to surveyed field boundaries or measured hydrology.

The preview follows the pinned HDRP LayeredLitData.hlsl ComputeMaskWeights top-down rule: each layer consumes the minimum of its channel and the remaining weight. Source albedos are decoded into linear light once, blended there, and encoded back for the review PNG. No runtime shader, tint, lighting or exposure modification is proposed.

The calculated result changes 481,391 R pixels. For the 341,716 pixels with eligibility above 0.5:

All G/B/A bytes are identical. Pixels outside eligibility and original woodland R are identical. The live summer mask, winter mask, DEM, route source and original source photographs remain unchanged and hash-checked. No additional layers, textures, geometry or runtime allocation are introduced by eventual mask publication.

The comparison was viewed and is suitable for the next limited native test: its greens are lower and its broad meadow coverage is less uniform without a brown blanket or pronounced grid. Native lighting, material mips, the near roadside grass and the transition into forest must still be assessed in the same Alps01 view0/view1. It is not an artistic pass. Root owns publication after inspecting the candidate and finishing the current Unity build; the existing live summer-mask generator must retain this frozen candidate composition if later regenerated.