Formalize depth layers with their own stone/wall types, cave density, ore tables, background tint and spawn tables. Add a strata.mapgen.layer_at(y) API that other systems (spawning, music, biome effects) can query. Tune depths so there are about 3 minutes of digging from the surface to the underworld.
Formalize depth layers with their own stone/wall types, cave density, ore tables, background tint and spawn tables. Add a `strata.mapgen.layer_at(y)` API that other systems (spawning, music, biome effects) can query. Tune depths so there are about 3 minutes of digging from the surface to the underworld.
claude
added this to the v0.3 World & Biomes milestone 2026-11-07 18:28:29 +00:00
The mapgen is driven by a layer table, strata.mapgen.layers. Each row has the layer's stone and wall, tunnel width, cavern threshold, ore table, background colour and spawn table. strata.mapgen.layer_at(y) returns the row for a height.
Stone gets harder: slate is cracky = 2 (slow with copper), basalt is cracky = 1 (needs an iron pickaxe).
Boundaries dither over 6 rows; cave sizes ease over 16 rows.
The creature spawner reads the layer's spawn table (strata.spawn_candidates). Only the surface has an entry so far (slimes by day).
Below the surface layer the player's sky becomes the layer's flat colour with no sun, moon, stars or clouds. Walls are tinted per layer through their own textures.
Slate and basalt walls can be crafted at the Work Bench.
Depth tuning. Real dig times are about 0.18 s per stone block with a copper pickaxe, much faster than I first assumed, so the world is deeper than planned. The selftest digs 11 straight shafts to the underworld: 673 blocks dug and 239 rows of open cave on average, 168 s with a copper pickaxe.
Tests. 13 new cases in tests/strata_selftest/tests/layers.lua. Full suite: 79/79 pass locally, zero ERROR lines.
Not checked in the client: how the layer colours and new textures look, and the sky change when crossing a layer.
Note for existing worlds: areas already generated keep their old caves, so there are seams where new chunks meet old ones.
Done in 0e0c431.
**What changed**
- The mapgen is driven by a layer table, `strata.mapgen.layers`. Each row has the layer's stone and wall, tunnel width, cavern threshold, ore table, background colour and spawn table. `strata.mapgen.layer_at(y)` returns the row for a height.
| Layer | From y | Stone / wall | Caves | Ores |
|---|---|---|---|---|
| surface | sky | stone | tunnels only | copper |
| underground | −20 | stone | tunnels, rare chambers (17% open) | iron, copper |
| cavern | −300 | slate (new) | large caverns (31% open) | iron (richer), copper |
| underworld | −900 | basalt (new) | mostly open (74%) | none yet (#25) |
- Unbreakable bedrock from y = −1000 down.
- Stone gets harder: slate is `cracky = 2` (slow with copper), basalt is `cracky = 1` (needs an iron pickaxe).
- Boundaries dither over 6 rows; cave sizes ease over 16 rows.
- The creature spawner reads the layer's spawn table (`strata.spawn_candidates`). Only the surface has an entry so far (slimes by day).
- Below the surface layer the player's sky becomes the layer's flat colour with no sun, moon, stars or clouds. Walls are tinted per layer through their own textures.
- Slate and basalt walls can be crafted at the Work Bench.
**Depth tuning.** Real dig times are about 0.18 s per stone block with a copper pickaxe, much faster than I first assumed, so the world is deeper than planned. The selftest digs 11 straight shafts to the underworld: 673 blocks dug and 239 rows of open cave on average, **168 s with a copper pickaxe**.
**Tests.** 13 new cases in `tests/strata_selftest/tests/layers.lua`. Full suite: 79/79 pass locally, zero `ERROR` lines.
**Not checked in the client:** how the layer colours and new textures look, and the sky change when crossing a layer.
**Note for existing worlds:** areas already generated keep their old caves, so there are seams where new chunks meet old ones.
CI follow-up: runs 19916 and 19917 failed on the commits above. Two of the new cases timed out, because the shared runner took 49.5 s to generate the deep test column against a 20 s case limit; no assertion failed. Fixed in 140c2e0 (per-case timeouts, a narrower column). Run 19918 is green.
CI follow-up: runs 19916 and 19917 failed on the commits above. Two of the new cases timed out, because the shared runner took 49.5 s to generate the deep test column against a 20 s case limit; no assertion failed. Fixed in 140c2e0 (per-case timeouts, a narrower column). Run [19918](https://gitea.eom.dev/Gaming/luanti_game_strata/actions/runs/19918) is green.
Blocking a user prevents them from interacting with repositories, such as opening or commenting on pull requests or issues. Learn more about blocking a user.
Formalize depth layers with their own stone/wall types, cave density, ore tables, background tint and spawn tables. Add a
strata.mapgen.layer_at(y)API that other systems (spawning, music, biome effects) can query. Tune depths so there are about 3 minutes of digging from the surface to the underworld.Done in
0e0c431.What changed
strata.mapgen.layers. Each row has the layer's stone and wall, tunnel width, cavern threshold, ore table, background colour and spawn table.strata.mapgen.layer_at(y)returns the row for a height.cracky = 2(slow with copper), basalt iscracky = 1(needs an iron pickaxe).strata.spawn_candidates). Only the surface has an entry so far (slimes by day).Depth tuning. Real dig times are about 0.18 s per stone block with a copper pickaxe, much faster than I first assumed, so the world is deeper than planned. The selftest digs 11 straight shafts to the underworld: 673 blocks dug and 239 rows of open cave on average, 168 s with a copper pickaxe.
Tests. 13 new cases in
tests/strata_selftest/tests/layers.lua. Full suite: 79/79 pass locally, zeroERRORlines.Not checked in the client: how the layer colours and new textures look, and the sky change when crossing a layer.
Note for existing worlds: areas already generated keep their old caves, so there are seams where new chunks meet old ones.
CI follow-up: runs 19916 and 19917 failed on the commits above. Two of the new cases timed out, because the shared runner took 49.5 s to generate the deep test column against a 20 s case limit; no assertion failed. Fixed in
140c2e0(per-case timeouts, a narrower column). Run 19918 is green.