Nesting Bitmap Polygon

Origin point for sheet placement

List of geometries to be nested. Any shape — curves, concavity, holes handled natively.

Width of the sheet used for nesting

Height of the sheet used for nesting

Spacing between sheets when laying them out

Distance from the edges of the sheet to the nested parts

Minimum clearance between parts (kerf). Dilates each part mask.

Discrete rotation poses per part. 1 = no rotation, 2 = fast and the best default, 4 and 8 sample finer. The arc they span is chosen per part: shapes that tile when flipped (wedges, triangles, L-brackets) get the full circle, shapes that nest by translation (arcs) get 0..π. More poses cost proportionally more time and rarely change the sheet count — raise only for strongly irregular parts.

Precision for discretizing curves to polylines (mm). Smaller = more samples per curve.

Sheet boundary curves. Rectangular outlines for each sheet used.

Nested geometries transformed to their final positions on sheets.

Translation vectors for each input geometry. Unplaced geometries have zero vectors.

Rotation angles in radians for each input geometry. Unplaced geometries have zero rotation.

Sheet index for each input geometry. Unplaced geometries have value -1.

Original input geometry indices for each nested geometry.

Fraction of usable sheet area covered by parts (0-1), per sheet. Edge clearance excluded.

Largest reusable empty rectangle per sheet (CAM remnant-stock detection).

One diagnostic string per unplaced part, or empty if all parts were placed.

BLF — fast greedy bottom-left | GA — genetic algorithm over order + rotation (best density, slow)

When all parts are uniform-radius circles, place analytically into a hex grid (skips BLF/GA). Falls back silently for non-circle input.

mm per pixel. Smaller = more accurate but slower + more memory. Auto-adapts upward for huge sheets (capped at 500px/side), so this only matters at face value for sheets ≤500mm/side.

Post-pass compaction iterations. 0 = disabled (default — benchmarked to add no density over the lowest-top placement while roughly doubling run time). Raise to trade speed for a possible tighter pack on hard inputs.

PRNG seed for GA. Same seed + input = bit-exact same output.

GA population size

GA maximum generations. Each generation re-runs the full placement popSize times, so keep it modest — the GA is seeded from the BLF result and only searches for improvements on top of it.

Stop GA after this many generations with no improvement. Low by default so runs where BLF is already optimal (the GA seed) finish quickly instead of grinding stagnant generations.

GA order-mutation rate

GA rotation-mutation rate