Watertight & Manifold

What does it mean for a model to be watertight or 'manifold'?

A model is watertight when its surface fully encloses a volume — if you poured water inside, none would leak out. There are no holes, no missing patches, no gaps where inside meets outside. This builds directly on the idea from mesh basics: a mesh is only a surface, so that surface has to close completely to describe a real solid.

Manifold is the stricter, more technical version of the same idea. A manifold mesh is one that could exist as a real, solid object. The key rule: every edge must be shared by exactly two faces — no more, no fewer. An edge with only one face means there is a hole. An edge shared by three or more faces means the surface branches in a way no solid object can.

In everyday use the two words are often mixed. Think of it this way: watertight describes the goal (a sealed volume), manifold describes the precise geometric condition that guarantees it.

Why must a model be watertight to print?

The slicer's whole job is to decide, for every point in space, whether it is inside the object (fill it with material) or outside (leave it empty). It works this out from the surface: cross a surface and you move from outside to inside.

A hole breaks this logic. If the surface is not closed, the slicer can no longer tell where inside ends and outside begins — the boundary leaks. The result is unpredictable: the slicer may refuse the model, silently 'repair' it in a way you did not intend, or produce strange missing or filled regions in the print.

This is why watertightness is not a nice-to-have — it is the minimum requirement for a model to be printable at all.

What are the common ways a mesh becomes non-manifold?

Holes (missing faces): the most obvious — a gap in the surface where a face is absent. The volume is no longer sealed.

Edges shared by more than two faces: for example an internal wall joining the outer surface, or two solids fused so a face passes through the interior. The surface branches, which no real solid does.

Internal or duplicate faces: leftover geometry hidden inside the model, or two copies of the same face stacked together. The printer sees contradictory information about what is solid.

Self-intersections: the surface passes through itself, so a region is both inside and outside at once.

Zero-thickness geometry: a single flat face or a loose edge with no volume behind it. It looks like a surface but encloses nothing — common when a wall is modelled with no thickness.

Why does code-based modeling usually stay watertight?

This is one of the quiet advantages of tools like OpenSCAD. When you build with solid primitives (`cube`, `cylinder`, `sphere`) and combine them with valid boolean operations (`union`, `difference`, `intersection`), the result of combining watertight solids is itself watertight. You rarely create holes by accident, because you are never editing individual faces.

Mesh editing in a tool like Blender is more powerful but easier to break: delete one face, extrude carelessly, or leave duplicate vertices, and the mesh is instantly non-manifold. This is exactly why, to deliberately create a broken mesh for testing, you reach for Blender — OpenSCAD resists making one.

There is one important catch, though. OpenSCAD can produce trouble when faces land in exactly the same place — for instance subtracting a hole whose top sits perfectly flush with the surface, leaving a zero-thickness coincident face. The fix is a habit you may already have seen: make the cutting shape slightly overlap (a millimetre longer, started a millimetre lower). That is why example code so often uses values like `h + 2` and `translate([0, 0, -1])` — the overlap guarantees a clean cut with no coincident faces.

How do you detect and fix non-manifold problems?

The slicer is your first alarm. When you import a model, slicer software typically detects non-manifold or non-watertight meshes and warns you — often offering to auto-repair. Auto-repair is a safety net, not a plan: it guesses, and the guess may not match your intent.

Dedicated repair tools can diagnose and fix more carefully — for example Blender's built-in 3D-Print toolbox, which highlights non-manifold edges and can patch holes.

The best fix is upstream: model so the problem never appears. Use solid booleans, give cutting shapes a small overlap, and avoid zero-thickness walls. A clean model needs no repair.

What about hollow parts — and why does this matter beyond plastic?

Hollowing a model to save material is common, but a hollow object must still be watertight: the inner and outer surfaces together have to seal properly. An open container — like a box without a lid — is fine, because it is genuinely open. A sealed hollow is where care is needed.

This becomes critical with other processes. In resin printing, a fully sealed hollow traps liquid resin and needs drain holes. In metal powder processes, a sealed hollow traps unfused powder that must escape. So 'watertight' sometimes has to be balanced against 'let the trapped stuff out' — a tension we return to in the process guides.