[raykholo]

Port of tangency-labs#13906 (Unfold for Printing / foldable 3D print, by @cvengels) onto the surfaces-first develop in tangency-labs#9245. Draft: the original was a WIP, and the gaps below are still open.

What it does

• Unfold for Printing wizard: splits a solid into flat printable parts that fold back up along film hinges. It has V-grooves, bevels, filament channels, press fits and carved hole walls.
• sheet.convert takes print_* parameters: wall, film, clearance, split edges, filament edges, press-fit edges, and swap start face.

How it maps onto the surfaces

• New SurfaceSheet::print_layout(&Development) (crates/doc/src/sheet_print_layout.rs):
  • Each piece of the development is a part.
  • Panels hang on each other along the bend joints.
  • Each child panel is translated so its sharp line lies on its parent's. That takes the zone width out, which is the hinge on the film at the outer skin.
• Rip cut-backs are set through the sheet's own Lap before develop():
  • butt joints: Cut(0, (t+c−min(t·cosφ,0))/sinφ)
  • mitres: Cut(s, s), with s = c/2/sin(half)
  • filament edges: Cut(0,0)
• Split edges are joins that stay cut open, so the faces past them become their own pieces.
• The sheet stays the definition: the flat is never written back. Unfold and Refold toggle between the flat print and the folded print. Other sheet features are refused on a print.

Commits

• The five original commits are cherry-picked with authorship kept. Their old-model hunks were dropped where Sheet metal: one model, the part's surfaces; the flat pattern is derived (#6748) tangency-labs/solidcraft#9245 deletes that code, and the wizard commit's body says how.
• 6f34cc7: the layout and the solid now come from the surface sheet.
• 14531f1: the tests are rewritten in closed forms. Old and new assertions:
  • v_grooves: V = t·(6000+2·100·30+2·60·30−4·30·(t+c)) − (2(100+60)−4(t+c))·groove
  • butt and mitre formulas for c = 0 and c = 0.2
  • filament: butt − five ≈ 4·30·1.85²
  • The rest are listed in the commit body.
• Hexagonal prism, every edge pressed: folded is 0.40 % below flat (23556.16 vs 23650.32 mm³). On the old model the loss was under 0.3 %. Folding can only lose the corner material that two sides' fingers share, so the test asserts folded < flat, with the loss under 0.42 %.

Open gaps

1. The wizard's interactive steps still use their own B-rep unfolder (UnfoldGraph/FlatLayout, generator 2's wall_solids). They run before any sheet exists. Generator 1 (sheet.convert) uses the surface sheet.
2. print_wall is always required: a shelled body's own walls are not read.
3. Hole mitres and sockets assume the hole wall is square to its face.
4. Corner fills and channels that run out through a folded-over face are not ported. The old corner-film assertions are dropped, not replaced.
5. Further gaps:
   • Curved or multi-point rip edges get no bevel or joints.
   • Inside corners use Miter{max(c, 0.8)}.
   • Convert's folded-overlap check is skipped for prints.
   • Cuts made after the convert are not guarded.
   • There is no print-specific DXF or bend-table handling.

Tests (buildq)

• engine foldable: 10 passed. engine print_unfold: 10 passed.
• engine convert_tests: all pass apart from two failures that already fail on fix-6748-develop.
• engine schema: 719 passed.
• ui-egui dialogs_sheet: 46 passed. ui-egui prefs: passes apart from the stand-up failure that also fails on the base.

Credit: @cvengels (C. Vengels) for the wizard, the print joints and the original PR.