Worldgen Phase 1: apply BiomeDefinition TerrainModifier during height shaping

[MichaelFisher1997]

Parent: #646

Phase

Phase 1: Terrain modifier integration

Goal

Wire BiomeDefinition.terrain into height generation so biome metadata has structural impact, not just surface/vegetation impact.

Tasks

• Decide where preliminary biome/family selection occurs before final height shaping.
• Apply height_amplitude, smoothing, clamp_to_sea_level, and height_offset in a deterministic order.
• Avoid circular instability where height changes cause biome flips causing height changes.
• Keep ocean/coast hard decisions stable.
• Add or update tests for modifier application.

Acceptance Criteria

• Swamp-like modifiers can flatten/clamp terrain near sea level.
• Mountain-like modifiers can amplify terrain without breaking peak compression.
• Existing generation remains deterministic for fixed seeds.
• nix develop --command zig build test passes.

Dependencies

• Should follow Phase 0 selector/modifier tests.

Parallelization

This is the main serial integration task for Phase 1. Tuning issues can run after this lands.

[github-actions]

Based on my analysis of the codebase, here's what I found:

Classification: Feature Request

Current Architecture Limitation:

The TerrainModifier fields (height_amplitude, smoothing, clamp_to_sea_level, height_offset) exist in BiomeDefinition at modules/world-worldgen/src/biome_registry.zig:56-65 but are not applied during height computation.

Key issue in modules/world-worldgen/src/terrain_shape_generator.zig:185-338:

1. Height is computed first via self.height_sampler.computeHeight(...) (line ~156)
2. Biome selection happens AFTER height is set (line ~265: self.biome_source.selectBiome(...))
3. The biome's terrain field is retrieved but only used for filler_depths and coastal_types - NOT for height modification

The height generation in modules/world-worldgen/src/height_sampler.zig:208-312 uses RegionControls (height_mult, drama_mask) from the region system, not per-biome TerrainModifier values.

Required Changes:

1. Preliminary biome/family selection must occur BEFORE final height shaping (or use a layered approach where biome modifiers are applied after base height)
2. The computeHeight() function needs to accept and apply TerrainModifier parameters
3. The deterministic order mentioned in the issue suggests: base height → biome modifier application → sea-level clamping

Implementation Approach:

• Option A: Two-pass height (base shape → biome selection → modifier application)
• Option B: Pass preliminary biome context into computeHeight() for modifier application
• Must preserve the ocean/coast hard decisions that happen in STEP 1 of computeHeight()

Risk: Circular instability (height changes → biome flips → height changes) needs careful handling. The issue notes suggest keeping biome selection stable once made.

This is well-scoped for Phase 1. The parent issue #646 tracks the full multi-phase effort.

opencode session  |  github run