Advanced mode

Panelize and fabricate

The end of the pipeline: turning something that exists in a viewport into something that exists on a workbench. Flat panels, interlocking ribs, unrolled strips, and sheets nested for a cutter.

Panelize

A curved surface made of quads has a problem: its quads are not flat. Four points in space are almost never coplanar, and a panel that is not flat cannot be cut from sheet glass, sheet metal or plywood without being formed — which is expensive enough to decide the project.

Planarise nudges every face onto its own best-fit plane while trying to hold the overall shape. These two goals conflict, and the controls are how you decide the trade.

Iterations: Number of planarisation passes.
Planarity: Strength pulling faces flat.
Shape: Strength holding edges near their original lengths. At zero the mesh is free to become whatever is most planar, which is frequently not the shape you designed.
Pin boundary: Holds the open boundary while planarising, so the outline you designed survives the optimisation.
Tolerance: Largest acceptable planarity deviation, as a fraction of panel size. This is where your fabricator's tolerance goes.

Analyse planarity

Analyse planarity measures every face and writes the deviation to an attribute, without changing anything. Turn on the planarity overlay and you get a heatmap with a legend: green panels are within tolerance, red ones are not.

Run this before planarising. It tells you whether you have a problem at all, and where. A single-curvature surface like a Vault is already planar everywhere and needs no optimisation; a Saddle is doubly curved and will fight you at every panel.

The geometric fact underneath

A doubly-curved surface cannot be built from flat quads without changing shape. That is not a limitation of this tool. Your options are: accept shape change, triangulate (three points are always coplanar), subdivide into smaller panels where deviation is small, or form the panels. The analysis tells you how much of each you need.

Fabrication modes

Contours

Parallel slices through the mesh, each becoming a closed profile. Stack the physical slices and you have the form. The oldest trick in model-making and still one of the best.

Axis: X, Y or Z.
Slices1 – 500: Number of slices taken across the mesh.
Use spacing: Place slices at a fixed distance apart instead of a fixed count. This is the one you want when the material thickness is what determines the spacing — which is always, if you intend to build it.
Spacing: Distance between slices when using spacing.

Waffle

Two sets of ribs crossing at right angles, each slotted so they interlock. Assembled, they hold the form of the surface with no glue and no fasteners. This is the eggcrate construction behind most laser-cut architectural models and a lot of furniture.

Ribs X / Ribs Y1 – 200 each: How many ribs run in each direction. More ribs means a more accurate form and more assembly.
Material: Thickness of the sheet the ribs are cut from. The slots are cut to this width, so getting it right is the difference between a kit that assembles and a pile of scrap. Measure your actual material — nominal 3mm ply is rarely 3mm.
Slot tolerance: Extra width added to each slot so the parts actually fit. Start small and cut a test joint. Too little and it will not go together; too much and it wobbles.

Unroll

Developable strip unrolling: takes a surface that can be flattened without stretching and lays it out flat, with a distortion readout so you know what the flattening cost.

A cylinder or a cone unrolls perfectly. A sphere does not — this is the same reason every world map is a lie. The distortion figure tells you which case you are in and how badly.

Kerf and nesting

Kerf: Width of material removed by the cutter, compensated for in the output. A laser removes roughly 0.1–0.2mm; a CNC router removes the full bit diameter. Uncompensated kerf means every part comes out one kerf-width undersized in every direction — which for a slotted assembly means nothing fits.
Sheet width / height: Dimensions of your stock. Parts are nested onto sheets of this size, and as many sheets are generated as needed.
Margin: Gap left between nested parts and the sheet edge. Keeps parts from being cut into each other, and leaves the clamping edge alone.

Export

Format: SVG, DXF, or both. SVG is right for laser cutters and for anything you want to open in Illustrator or Inkscape. DXF is right for CNC and CAD, and is readable by essentially every CAM tool.
Path: Folder and base filename. // at the start means relative to the .blend file, which is worth using.
Scale: Multiplier converting scene units to millimetres. Default is 1000, which is correct if one Blender unit is one metre. Get this wrong and you will cut something a thousand times too small, which the material will not survive.
Label parts: Writes a panel number onto each exported part as a text entity. Assembling forty unlabelled ribs is a genuinely miserable afternoon.

Cut one test piece first

Before committing a full sheet: export a single slotted joint, cut it, and try to assemble it. Material thickness, kerf and slot tolerance interact, and one small test resolves all three at once. Every experienced maker does this and it is still the most-skipped step.

Before you send it to the cutter

  ☐  Scene units correct, export scale matches
  ☐  Material thickness measured, not assumed
  ☐  Kerf measured for this machine and this material
  ☐  Slot tolerance validated on a test joint
  ☐  Parts labelled
  ☐  Nesting fits the actual sheet you own
  ☐  Planarity within your fabricator's tolerance
  ☐  Part count sane — check the result line

The full pipeline

Everything in Rhea points at this. A complete run, using nothing else:

Start with a Saddle.
Give it structure with the Diagrid effect, and apply it.
Form-find it into a shape that carries its own load.
Analyse planarity. Planarise. Analyse again.
Generate waffle ribs with your real material thickness.
Nest, label, export SVG.
Cut it. Assemble it. It stands up.