Skip to content

feat(refine): only a thicket is heavily obscured - #3

Merged
GentBajko merged 3 commits into
mainfrom
feat/thicket-obscurement
Nov 15, 2026
Merged

GentBajko merged 3 commits into
mainfrom
feat/thicket-obscurement

Conversation

@GentBajko

Copy link
Copy Markdown
Member

For DiceMaster. Its ruling of 2026-11-14 keeps the SRD rule that a heavily obscured square blocks sight, and asks Arda to mark only thickets heavy. Until now Arda made a dense wood almost wholly heavy: four large trunks within 2.4 squares blocked sight, so a creature inside saw about 2 % of the squares within 60 ft.

The rule (logic/09 §rules-sidecar)

  • Thicket, heavily obscured (blocks_sight, scene obscured: heavy): a dry square with 4 or more pieces of woody understory anchored within 2 squares of its centre. Woody understory means small or stunted trees and woody shrubs (bush, juniper, krummholz, sagebrush); ferns don't count.
  • Wood floor, lightly obscured (lightly_obscured, scene obscured: light): a dry square with 2 or more large trunks within 2.4 squares, however closed the canopy. A reed bed counts as well.
  • Clear: anything else, unless a prop, wall or rock says otherwise. The catalogue's own sight blockers are untouched.
  • Why placed items: they are what the picture shows and what cover reads, so the rule is a pure function of them (output::sight in crates/arda-refine/src/output.rs). The scatter's hard-core spacing (grid 1.6–1.8 squares) makes 4 within 2 squares a local clump. Thresholds were tuned on 14 cells of the kit world.

Before and after (MICRO-42 kit world, scene at 64 px per square)

Visibility is measured from the cell's centre: the share of squares within 60 ft whose line crosses no heavy square.

Cell Kind heavy before → after light after Visible within 60 ft, before → after
479, 1245 (kit tactical-cell) dense wood 90.7 % → 0.2 % 99.8 % 8/440 (1.8 %) → 440/440
632, 1160 dense wood 88.8 % → 0.0 % 99.9 % 1.8 % → 100 %
560, 776 dense wood 90.0 % → 0.0 % 99.8 % 1.8 % → 100 %
560, 1160 dense wood 86.8 % → 1.1 % 97.4 % 1.8 % → 82.5 %
584, 632 dense wood 70.2 % → 0.0 % 91.5 % 1.8 % → 100 %
632, 1352 mid wood 11.8 % → 2.8 % 41.5 % 60.7 % → 74.8 %
1016, 1256 scrub 19.2 % → 1.1 % 33.1 % 100 % → 100 %
752, 1880 scrub 23.6 % → 1.3 % 39.6 % 100 % → 98.6 %

Contract kit

  • Re-recorded: the tactical-cell, tactical-scene-day and tactical-scene-night goldens, and the tactical-window body hash.
  • Unchanged: every image (PNG and tiles), every dungeon and site case and every other body.

Evidence

  • Unit tests (output::tests):
    • a closed canopy is light;
    • a thicket is heavy;
    • three shrubs, or any number of ferns, make no thicket;
    • understory beyond 2 squares does not count;
    • water is never a thicket, and reeds are light;
    • open ground is clear.
  • Gates: cargo fmt --check and cargo clippy --workspace --all-targets -D warnings pass. cargo test --workspace --all-features --release passes (contract kit, world_api, refine, blocks and scene suites included).

Docs

logic/09 (both rows and the thicket definition), API.md (scene obscured), the arda-refine README, CHANGELOG.md (Unreleased, flagged as a content change) and the ledger.

Process note

The user was away, so spec and plan were written from the DiceMaster session's brief without the approval gates; this PR is the gate. CHANGELOG.md and docs/capstone/changelog.md will need a trivial conflict resolution against #2 and the memory-ceiling PR.

@GentBajko

Copy link
Copy Markdown
Member Author

Converted to draft: Arda's feature process needs the maintainer's approval of this change's plan before code, and that approval has not been given. The code here was written ahead of it. The maintainer can approve the plan as built (review this PR and mark it ready) or ask for the plan to be redone first.

@GentBajko
GentBajko merged commit 4f5ede7 into main Nov 15, 2026
14 checks passed
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

1 participant