Quality & mesh

• Draft (Free+) — the coarsest mesh. Shape, mass and deflection are already right here (within 1% of the finer tiers). No safety factor is quoted at Draft.
• Standard (Maker+) — everyday studies: the shape and its deflection, converged to within 1% of High. No safety factor is quoted at Standard either.
• High (Workshop+) — the tier a stress verdict is quoted from: the validation mesh is refined at the hot spot to the element size the critical-distance verdict needs. Higher credits.

Finer meshes than the tier default raise credits roughly with 1/size³. See Subscription for plan unlocks.

Draft is a known offset from Standard, and the offset has been measured — nine surface-stress pairs across eight parts, same loads, same supports, same solver.

Deflection, stiffness, mass and the delivered shape are right.Across all eight parts Draft landed within 1.1% of Standard, and it never read softer than Standard. If your question is “how much does it bend”, “how much lighter did it get” or “what shape does this become”, Draft answers it and you do not need to pay for Standard to find out.

Peak stress is the exception, and it is not one number off. Draft read 4–18% belowStandard’s raw peak, in two distinct groups:

• About 4–5% lowwhere the peak sits on a real feature of the part — a fillet, a wall, a load path. That peak converges: refine the mesh and it settles down.
• 10–18% lowwhere the peak sits on a clamp, a fully fixed face, a bore edge or a sharp internal corner. That peak never converges — on one bracket it went 21 → 24 → 26 MPa across Draft, Standard and High and was still climbing. In that case Standard is not the right answer either; it is a finer arbitrary one, and the honest reading is that the geometry (or the support) needs a radius, not a finer mesh.

And against a verdict, the sign flips.A safety factor is not read off the raw peak — it comes from a critical-distance average at the hot spot, which sits below the peak. Measured against that, Draft’s peak was 22–24% high: pessimistic, not optimistic.

One answer, on every tier and every plan: a body-fitted TET10 mesh — curved ten-node tetrahedra generated from your CAD, so a fillet is a fillet rather than a staircase of cubes. There is no mesh choice to make and no mesh setting behind a plan.

Older material on this site described a Cartesian Grid (HEX8), a Fine tet upgrade, and an Enterprise HEXmesh. Those design-grid backends still exist inside the engine, and internal validation runs still drive them, but the product no longer offers any of them — a study that asks for one is refused rather than quietly given something else. If you read that here before, it is gone, not hidden.

After topology, Exoform extracts a printable isosurface (marching cubes). On Draft it upsamples the density field, Gaussian-prefilters, then Taubin-smooths the mesh — seconds of CPU, no extra credits, and not a TET10 remesh. Raise Smooth organic mesh in Setup for softer walls; set 0 to keep sharp preserved holes.

Prove Supports/Loads on Draft → then raise quality once displacement direction and order-of-magnitude stress look right. A beautiful High mesh of a wrong load case is still wrong.

Liked a Draft topology direction? On the study detail page use Refine further → Standard/High — it prolongates the prior density onto the finer grid (cheaper than a cold re-solve). FEA studies get Re-run analysis at the finer mesh.

Next: Topology · Results.