Fix: small covered regions (e.g. lego studs) not detected as fully covered on transition layers

[Copilot]

• Understand the issue: stud circles (~3-5% of brick area) fail the 10% minimum size ratio threshold in findCoveredRegions
• Fix cavity.coffee: remove the 10% minimum size ratio lower bound; keep 55% upper bound and all other guards
• Add test for "multiple small covered regions above large region" (lego stud scenario) in cavity.test.coffee
• Regenerate lego-brick.gcode benchmark — layers 151-154 now show SKIN (outer ring) + FILL (under each stud) instead of solid SKIN
• All 16 cavity tests pass (15 existing + 1 new)
• Fix cavity.coffee comment: "its area" → "the candidate region's area" (reviewer feedback)
• Fix test comment: area ratio ~2% not ~3-5% (reviewer feedback)
• Add explicit setNozzleDiameter(0.4) and setExposureDetectionResolution(961) to test (reviewer feedback)
• Strengthen fill assertion: count distinct fillRegions and assert ≥6 (one per stud) (reviewer feedback)

---

🔒 GitHub Advanced Security automatically protects Copilot coding agent pull requests. You can protect all pull requests by enabling Advanced Security for your repositories. Learn more about Advanced Security.

[Copilot]

Pull request overview

Fixes exposure-detection behavior on transition layers where small covered regions (e.g., lego studs) were not being recognized as “fully covered,” causing those areas to get skin infill instead of regular infill.

Changes:

• Remove the 10% minimum size-ratio guard in findCoveredRegions, allowing legitimately small covered features to be detected.
• Add a regression test covering the “small cylinders on a slab” scenario and asserting TYPE: SKIN + TYPE: FILL behavior on a transition layer.
• Regenerate the lego-brick.gcode benchmark artifact (LFS pointer update).

Reviewed changes

Copilot reviewed 3 out of 3 changed files in this pull request and generated 4 comments.

File	Description
src/slicer/skin/exposure/cavity.coffee	Updates covered-region detection heuristic by removing the minimum size-ratio bound and documenting the rationale.
src/slicer/skin/exposure/cavity.test.coffee	Adds a lego-stud-style regression test intended to ensure small covered features produce regular infill under them on transition layers.
resources/gcode/benchmarks/lego-brick.gcode	Updates the stored benchmark output (Git LFS pointer) to reflect new slicing behavior.

[jgphilpott]

@copilot apply changes based on the comments in this thread

[Copilot]

@copilot apply changes based on the comments in this thread

Applied all four review suggestions in commit 3660889:

• cavity.coffee line 9: "its area" → "the candidate region's area"
• Test comment line 1161: area ratio corrected to "~2%"
• Test setup: added explicit setNozzleDiameter(0.4) and setExposureDetectionResolution(961) calls
• Assertion strengthened: now tracks fillRegions (distinct consecutive FILL sections) and asserts fillRegions >= 6 — one region per stud