How to fix a Meshy model for 3D printing

Updated July 2026 · 5 min read

The generation came back looking great. Clean silhouette, convincing texture, the kind of thing you would happily show someone. Then you dropped it into your slicer, and the slicer either refused it outright or printed a hollow, shredded impression of the object you thought you had.

Nothing malfunctioned, and — worth saying, because an earlier version of this guide did not — Meshy has since built print tools of its own. The question is whether they ran on the file you are holding, and whether anyone checked.

A render and a slice ask different questions

A renderer and a slicer ask a 3D file two very different questions, and only one of them is easy.

The renderer asks: which triangles face the camera, and what colour are they? It draws the skin it can see and never looks behind it. A mesh can have holes in the back of the head, faces wound inside-out, and loose scraps of geometry rattling around inside the torso — and the render still comes out flawless, because none of that is visible from outside.

The slicer asks the hard question: for every point on every layer, is this inside the object or outside it? That only has an answer if the surface actually closes. Poke one hole in it and "inside" stops being defined. The model that looked finished is, to a slicer, not an object at all — just a shell of triangles floating in space.

The reason is in the training. Image-to-3D and text-to-3D systems are optimised against visual objectives: the reward is that the renders match the target from a set of viewpoints. Watertightness is not in the loss function. It is never even measured. On top of that, the mesh you download is usually extracted from an implicit representation — a density field or an SDF — turned into triangles and then decimated to keep the file small. Every step in that chain can leave a hole, invert a winding, or strand a fragment of geometry inside the body. It is the same failure pattern behind every AI-generated model that won't print.

Texture is not geometry — and STL keeps neither

Here is the assumption that costs people an evening: a good-looking texture means a good mesh underneath. It does not. Colour is painted onto the surface. It says nothing whatsoever about whether that surface encloses a volume.

Multi-view, textured generation makes this worse, not better. When a system is reconciling several viewpoints into one body, it can leave behind internal faces — surfaces that sit inside the model where you will never see them — and small floating fragments sealed within the torso like swallowed marbles. From the outside, invisible. To the slicer, more surfaces to reason about, more "inside" boundaries, more contradictory answers per layer. That is where the mystery warnings and the bizarre infill come from.

So you export to STL, and the first surprise lands: STL carries no colour at all. None. It stores triangles and nothing else — no texture, no UVs, no materials, no vertex colours. Going from a textured GLB or OBJ to an STL throws the entire appearance away. That is not a broken export. That is what STL is.

What survives is the mesh — and the mesh is the part that has to be watertight. STL keeps no colour, and it keeps no promises about the geometry either. It hands the slicer a pile of triangles and lets the slicer find out.

What actually breaks

The same faults show up again and again — and they are not all the same kind of problem.

The first four are plumbing: the mesh describes the right shape badly, and a repair tool fixes that without touching your design. The last two are not repairs at all. A thin wall is the shape being wrong. Scale is a slider in your slicer.

Use Meshy's own print tools first

Meshy documents a 3D-printing workflow, and you should use it before you use anything else. As of July 2026 it describes a built-in printability check that it says detects and repairs non-manifold edges, degenerate faces, holes and open boundaries, plus a remesh step. If the platform that made the mesh offers to fix the mesh, start there. Repairing a wreck downstream is the long way round.

It also gives you a real format decision, and this is where its own documentation is worth reading twice: STL carries no colour at all, so it is the single-colour route. 3MF is the one that keeps colour, for multi-material and colour printing. Meshy's own docs note that 3MF colour mapping can come out wrong in the slicer and tell you to check it there. So pick the format for the print you actually want — and if colour matters, do not export STL and wonder where it went.

Then verify — because a check that ran is not a check on your file

Meshy's printability check is Meshy's statement about its own output. It may well be right. But it is not a verification of the file sitting in your downloads folder, and you may not have run it at all — plenty of models come out of the default flow, or the API, or from someone else entirely.

SliceReady is the second opinion, and it runs on your machine with WebAssembly. The file is never uploaded — there is no server to upload it to.

1. Run Meshy's printability check and remesh if you have not. Then export: STL for a single-colour print, 3MF if you need the colour.
2. Drop the file onto SliceReady.
3. The damage is counted and highlighted on the model: open edges, flipped normals, degenerate triangles.
4. The repair runs and gives you a plain-language ledger of exactly what changed — how many holes were closed, how many normals corrected, how many dead triangles removed.
5. Download the repaired STL and slice it.

The one it can't fix: thin walls

This is the fault that catches the people who did everything else right. The mesh is watertight, the normals agree, the slicer accepts it without a murmur — and the sword blade and the outstretched fingers come out as ragged gaps, or simply never appear.

The reason is physical, not digital. A typical nozzle is 0.4 mm across. It cannot lay down a wall thinner than that, and a generator that never once thought about extrusion width will happily hand you a cape 0.1 mm thick. The geometry is valid. There is just nothing there to print.

Character prompts are the worst offenders, and it is always the same parts: blades, fingers, hair, capes. Thicken them in a modelling tool, or scale the whole model up until the thin bits clear your nozzle width. Knowing which of the two problems you have — plumbing or shape — is what saves the evening.

Frequently asked questions