Walls, perimeters and infill
You print a part at 20% infill, pick it up, and the inside is nearly hollow: the printer traces a few outer loops and packs the cavity with a sparse lattice that's mostly air. Here's the trap almost everyone falls into: when a part flexes or snaps, the first instinct is to crank up the infill. But that's almost never where the strength is. Strength lives in the walls — in the continuous beads that loop around the contour. Infill does something else, something important but different, and confusing the two costs you hours of print time and spools of plastic without the part holding up any better.
A part is a shell, not a solid block
Look at any print edge-on. The outer surface is made of a small number of loops, one nested against the next. Each loop is one pass of the nozzle: a perimeter. Two, three, or four perimeters make up the wall. The top and bottom are capped with a few solid layers — the top and bottom — and everything in between is infill: a mesh that crosses the interior, leaving gaps on purpose.
Those three things aren't interchangeable. Perimeters are continuous beads that run the full contour without a break. Infill is a mesh of crossing strands with gaps between them. That difference decides almost everything that follows: a continuous bead and a gapped mesh don't carry load the same way.
Perimeters carry the strength, not the infill
When you flex a part, the outermost material works hardest: peak stress sits at the outer face — the one that stretches on one side and compresses on the other — exactly where the perimeters are. A bead that loops all the way around carries load along its entire length, like a continuous fiber. Infill, by contrast, is a gapped mesh: it works far less and in a strongly direction-dependent way. Lines aligned with the load carry it; the transverse direction barely contributes, and the crossing junctions are the weak points that fail first. Under pure tension the stress does spread across the whole section, but even there each perimeter adds more continuous material than any reasonable infill percentage.
That's why going from 20% to 40% infill nearly doubles your print time and plastic, yet adds far less than double the strength. You're bulking up the core, the part that's stressed least. Above 30–40% the returns flatten: each added point of density costs the same and buys less.
The real lever is the perimeter count. Going from 2 to 3 walls thickens the entire shell, right where the part fights the load. On parts with internal volume, that extra perimeter usually beats a big jump in infill. On very thin enclosures — lots of wall surface, little hollow inside — it can cost more time, so measure it for your case. At a 0.4 mm bead width, 3 or 4 perimeters (≈1.2–1.6 mm of continuous wall) already form a continuous wall that works like a closed ring.
What infill is actually for
Infill doesn't raise strength proportionally, but it isn't useless either: it does three concrete jobs.
The first, and the one you can almost never skip, is to support the solid tops. The solid top layers aren't laid over air: they bridge across the infill mesh. With too little infill underneath, the top beads span gaps too wide for them, sag, and the surface dips — that's pillowing — the valleys and bumps you see on the top face. The infill is the scaffold the top closes over, which is why infill density and the number of top layers are decided together, not separately.
The second is stiffness against compression and buckling. A hollow shell handles the tension in its walls well, but under a load that crushes it or a wide panel liable to bow inward, the infill braces the walls against each other and stops them buckling. It's the bracing that keeps the shell from collapsing like an empty can.
The third is to spread local loads. A screw biting down, a boss, an impact zone — the infill behind it distributes that point force out to the walls instead of letting it concentrate and punch through the shell.
For those three jobs, 15–20% is enough on most parts, and 25–40% covers moderate load or impact. The pattern matters too, but with a caveat: grid is fast and reasonably uniform in the XY plane, while gyroid spreads stiffness more evenly across all three directions, at the cost of slower printing. No pattern fixes the truly weak axis — Z — because that weakness comes from the weld between layers, not from the infill pattern. Choose by what the part will actually take, not by default.
Size the wall in whole bead widths
This is the highest-payoff wall rule, and the cheapest to apply: make the wall thickness a whole multiple of the bead width set in your slicer.
The slicer can only fill a wall with complete beads. The number that governs this isn't the nozzle diameter; it's the tuned extrusion width: with a 0.4 mm nozzle the default usually sits around 0.45 mm and can be set from ~0.4 to ~0.6 mm. Check what value you have and do the math with that one. Say you leave it at 0.4 mm: if you draw a 1.0 mm wall, the slicer fits two perimeters (0.8 mm) and is left with a 0.2 mm strip it can't fill with a whole bead. Then it does one of two things, both bad: it leaves a thin gap down the center of the wall, which is exactly where it splits, or it overlaps the beads and bulges, deforming the face. Either way, you've built a worse wall by picking the wrong number.
Draw that wall at a clean multiple of the bead you use — 0.8 / 1.2 / 1.6 mm with a 0.4 bead — and every perimeter fits whole, with no leftover strip and no overlap. The wall comes out stronger and the surface cleaner, at no extra cost.
| Perimeters | Wall thickness | Use it for |
|---|---|---|
| 1 | ~0.4 mm | nothing structural — too brittle |
| 2 | ~0.8 mm | light parts, enclosures, brackets |
| 3 | ~1.2 mm | most functional parts |
| 4 | ~1.6 mm | parts that take real load |
Tops and bottoms: enough layers to close them
The same whole-multiple reasoning applies horizontally, but here the problem isn't the line thickness, it's the solid thickness that closes over the infill.
A single top layer doesn't close: it prints across the infill gaps and, with nothing under it on those spans, sags. The top face ends up with the infill showing through, rippled, dipped from cell to cell — the same pillowing as before — here caused by sparse infill, too few top layers, or weak cooling. That's why most slicers work in terms of a closed thickness, on the order of 1 mm, rather than a fixed layer count: at a 0.2 mm layer that's 4 or 5 solid layers, but at 0.1 mm you'd need twice as many for the same closure. The wider the infill cells, the more layers it takes to cover them without the pattern showing through.
It's the other side of the same trade-off: denser infill gives smaller cells and tops that are easier to close; sparse infill forces more solid thickness on top. Don't optimize one while ignoring the other.
Put it all together and a strong, efficient part has a clear recipe: walls sized to whole perimeters that take the load, a modest infill just dense enough to support the top and brace the walls, and tops and bottoms with enough thickness to close without dipping. Solid where it matters, hollow where it doesn't. If your goal is the opposite — shedding mass without losing stiffness — Lightweighting is where to go next. When you want to reinforce one specific area without bulking up the whole part, Ribs, gussets and fillets takes you from the wall to the geometry that spreads the load. All of this rests on how the beads weld to each other — even a four-perimeter wall delaminates if they don't fuse well — which is what Layer adhesion and anisotropy explains.