 Emil
|
22de8bc42e
|
fix: wider surface terrain — lower noise frequencies for rolling hills
base freq 0.08→0.012 (~260 cell wide hills, was ~40)
detail freq 0.23→0.04 (~78 cell features, was ~14)
micro freq 0.57→0.11 (~28 cell ripples, was ~6)
amplitudes reduced: base ±20→±15, detail ±10→±7.5, micro ±5→±2.5
Result: wide rolling hills instead of narrow sharp peaks.
|
2026-06-22 17:05:38 +03:00 |
|
|
Emil
|
4c9a074bb0
|
feat: world scale ×5 — trees, pools, walls, rooms, corridors all 5× larger
- WORLD_SCALE=5 constant in worldgen.rs, adjustable via single constant
- Trees: trunk 30 cells tall, canopy radius 10 (was 6/2)
- Pools: radius 10-30 (was 3-6)
- Walls: height 10-35, width 5 (was 3-7/2)
- Sand dunes: width 20-80 (was 6-16)
- BSP rooms: min 20×20, corridors width 5 (was 9×9/2)
- Surface terrain: amplitude ×5 (±20 vs ±4)
- Dirt layer depth: 40 cells (was 8)
- Spawn offsets ×5 for goblins/slimes
- Dirty rects cleared after world generation to avoid CA processing static terrain
- Chunk activation by dirty rects only (not modified flag)
- Dirty rect size limit 4096 cells to prevent CA spikes
- CA fill_prob for caves reduced 0.45→0.38 for larger open spaces
- All 185 tests + 14 scenarios pass
- Benchmark: ~18 FPS (regression from 85 FPS due to larger world features)
- Performance optimization deferred to separate pass
|
2026-06-22 08:27:15 +03:00 |
|