Fillet or chamfer: which edges to round, which to cut at 45 degrees, and why the bottom one is different

18 September 2026

geometrybasics

Chamfer the bottom edge, round the vertical corners, round every inside corner, and on a top edge take whichever you like the look of. That is the whole rule. It is not a matter of taste, because a printer builds a part upward in flat layers off the plate, and the shape of an edge decides how far each layer has to reach out over the one below it. A fillet and a chamfer of the same size are completely different propositions depending on which way the edge faces.

The bottom edge: chamfer

Put a 2 mm fillet on the bottom outside edge of a box and slice it at 0.2 mm. The arc leaves the build plate horizontally, so the first layer is the narrowest ring of the whole part and the second one has to reach 0.87 mm further out than it. That is 77 degrees off vertical. Nearly a millimetre of plastic goes down on air, on the layer with the least support under it. It curls, the fan cannot keep up, and the bottom of the part comes out chewed.

The same edge as a 45 degree chamfer steps outward by exactly one layer height per layer, 0.2 mm on 0.2 mm. That is the angle every slicer default, every fan duct and every bit of overhang advice is built around, and it prints without thinking about it.

A bottom chamfer earns its place twice, because it also swallows elephant foot. The first layer is squashed into the bed and spreads the outline by a few tenths of a millimetre; on a square edge that shows up as a lip you can catch a thumbnail on, and on a chamfer it just fills in some of the cut. The box generator defaults its edge chamfer to 1.2 mm for that reason, on the bottom edge of the base. It puts the same 1.2 mm on the top edge of the lid, where the print does not care at all and your hand does.

The top edge: either one, and printability does not decide

Going upward, a top edge makes the part narrower, so nothing overhangs anything and both shapes print. Here it really is about the look, and the surprise is that the fillet is often the worse looking one.

Near the top face a fillet's surface runs almost parallel to the layers, so each layer steps out by a lot and the curve reads as three or four visible terraces instead of an arc. A chamfer holds one constant step all the way along and comes out even. Below about 1.5 mm of radius I would take the chamfer on a top edge, which is the opposite of how the same two edges look on screen.

Vertical corners: round them, every time

A vertical edge is free to round. Every layer is the same outline, so there is no overhang anywhere and no cost to pay. What you get back is two things.

• The nozzle no longer has to stop dead and turn 90 degrees. That is where corner bulge comes from, and the ringing that follows it down the next wall.
• The part loses a stress riser at the exact spot a dropped box lands on.

The box generator's corner radius does both, and the corner bumpers ride on the arc it makes, so the radius is a structural choice there rather than a decorative one.

Inside corners: the fillet that actually holds

Everything above is about outside edges, where a fillet is mostly about how the part feels in the hand. The fillet that decides whether a part survives is the concave one: the inside corner where a wall meets a floor, a rib meets a face, or an arm meets the plate it is bolted to.

A sharp inside corner concentrates the whole load on a single line. On an FDM part that line almost always runs along a layer boundary, which is the weakest direction the part has, so the crack starts there and the arm snaps off clean. A fillet spreads the same load over an arc instead, and it adds material exactly where the bending moment is highest.

The hook generator says as much on its own tooltip: the rounding in the inside corner where the arm meets the plate is the single thing that most decides whether a printed hook holds. It suggests 5 mm, allows up to 12, and bigger genuinely does hold better.

Start from half the wall thickness. Then go bigger anywhere nothing has to sit flat in that corner. This is the one place where I would rather overdo it than get it exactly right.

What the printer can actually reproduce

A 0.4 mm nozzle lays a bead about 0.42 mm wide. Any fillet or chamfer smaller than one bead is a number in a file and nothing on the part: the slicer has nowhere to put it. Half a millimetre is the floor, 1 mm is where you can see it and feel it with a finger, and under 0.5 mm you are only paying for triangles in the STL.

Why the standards landed on 45 degrees

This is not just my habit. Gridfinity specifies every chamfer in the foot and the stacking lip at 45 degrees, which is why bins and baseplates print flat on the bed with no supports and still mate with somebody else's print from a different machine. The pegboard generator exports its connectors face down for the same reason: they point up as printed, with 45 degree chamfers, and never need a support.

45 degrees is the one angle that stays printable whichever way the part ends up on the plate. A standard that wants to be printed by strangers cannot afford anything else.

Where this is going

The browser CAD editor I have been writing about takes both, one edge at a time: the top rim rounded, the base rim chamfered, the vertical corner under them rounded again at its own radius. The piece I added this week is the inside one, the fillet at the root of a boss where it meets the wall it grew from, which until now was a field in the panel that quietly did nothing at all. That is the fillet from the hook, generalised, and it is the one I would not ship a load bearing part without.