Lightweighting
Every gram you print costs machine time, filament, and—on anything that moves—inertia you have to accelerate and brake a thousand times over. The temptation is to drop the infill and call it done. But infill is almost never where the weight that matters lives, nor where the stiffness you need comes from. Lightweighting well isn't blindly gouging out material: it's knowing which material does work and which only takes up space. And it turns out those two things sit in very specific places on the part—almost always the opposite ones from where you'd guess.
In bending, the material that works sits far from the center
Start by understanding what a part does when you load it. Most parts that fail do so in bending: you flex them, and they either deflect too much or snap. When you bend a bar, one face stretches and the opposite face compresses. Between them lies a plane that neither stretches nor compresses: the neutral axis. Here's the key to lightweighting: material hugging the neutral axis barely works, because it barely strains; the material that really works is the one far from the axis, in the outer fibers, where tension and compression peak.
That means the center of a solid section is almost dead weight. The part carries it, you pay for it in filament and time, and it adds next to nothing to bending stiffness. The design takeaway is direct: hollow out the center and leave the material at the perimeter. An I-section—two flanges separated by a thin web—or a box section—a thin-walled tube—puts almost all the plastic where the fibers work and clears the rest out of the middle. A well-proportioned I-beam can carry a fraction of the mass of the equivalent solid bar at nearly the same bending stiffness.
Two effects are worth committing to memory, because they're easy to confuse. First, in a solid section, bending stiffness scales with the cube of the depth. Double the depth of a solid bar in the direction of the load and it gets eight times stiffer. Second, once you hollow the section and leave the material in flanges or walls, what rules is no longer the cube of total depth, but how far you push that material from the neutral axis, and that scales with the square of the distance. Spread the flanges twice as far apart and their contribution multiplies by four, not eight. They're two distinct effects: raising the depth of a solid goes with the cube; pushing the flanges of a hollow profile farther out goes with the square. The moral is the same—pull material from the center and put it at the edge—but don't mix up the two laws.
Walls and ribs, not infill
In FDM that load-bearing shell has a name: the perimeters, the beads that loop around the contour. Almost all the stiffness of a printed part lives in them, not in the infill at the core. So the first lightweighting move is counterintuitive: add a perimeter and drop the infill. On a non-trivial part, a part with three walls at 10% infill is usually stiffer than the same part with two walls at 40%, and depending on size, also lighter: on small parts, that third perimeter can weigh as much as the infill you save, so check the actual volume before you assume you've saved weight. What is always true is that you move plastic from the dead center to the shell, which is exactly the perimeter that works hardest. Walls, perimeters and infill covers this in detail.
Scattered infill distributes stiffness poorly, for two reasons. First, it sits near the neutral axis, where it carries little load. Second, it isn't continuous material but a mesh of crossed strands with gaps, welded only where they cross. Per gram, that performs far worse in tension and bending than a continuous bead. Raising infill from 20 to 60% adds a lot of mass and a lot of time in exchange for a modest gain in stiffness. So when a lightened part keeps flexing, don't reach for the infill again: add a rib.
A rib is exactly the I-section idea brought down to a concrete part: a thin web stood on edge that raises the local section height and raises stiffness sharply. A rib is worth far more than raising infill from 10 to 50% across the whole part, for a fraction of the plastic and the time. Give it a thickness that's a whole multiple of the bead width—0.8 mm is two 0.4 lines; 1.2 mm, three—so the slicer fills it with clean perimeters and doesn't leave a thin, badly filled gap. And don't make it as slender as you'd like: a very tall, thin rib printed upright stops failing by strength and starts failing by buckling—it bows sideways under compression—on top of peeling off the bed during printing. Ribs, gussets and fillets covers how to size and place them, and how to keep them from concentrating stress at the root.
Hollow where stress is low: shells, lattices and organic shapes
A "solid" body in your model is solid because you drew it that way, not because physics demands it. Shelling—hollowing the interior and leaving a uniform-thickness skin—removes the dead mass from the center and keeps the load-bearing skin. A 30 mm solid block shelled to a 2 mm wall loses most of its weight and almost none of its bending stiffness, because the outer fibers are still there doing the work. Leave the shell at a whole number of bead widths thick so it prints as clean perimeters: with a 0.4 mm bead, 2 mm is exactly five perimeters; with a 0.45 mm bead, that same thickness leaves half a perimeter loose that the slicer fills badly. Tie the thickness to your actual line width, not to a round number.
When you shell you have to watch two more things that only show up once you drop the infill. First, if too little infill sits under the top face, the top layers find nothing to rest on and sag—pillowing—so add top layers when you drop the infill, not just perimeters. Second, a sealed internal cavity leaves a pocket of air. The slicer will leave it hollow either way, whether you model the cavity or not—it won't fill a cavity just because it's sealed—but that sealed pocket pressurizes as it prints and traps moisture. Give it a small drain or vent hole so it isn't sealed under pressure.
From there, the general idea is to remove material where stress is low and leave it where it's high. A lattice—a grid of struts or a repeated pattern of holes—replaces a full wall with a mesh that keeps the load paths and removes the rest. The organic, branching shapes you see in an optimized bracket are the same idea taken to the limit: material only along the paths the load flows through, from one anchor to another, and air everywhere else. It's structural, not decorative: material that doesn't fall on the path of the force adds nothing, so it genuinely is surplus.
This ties into how you choose the infill pattern when you do need internal support for the layers above. They don't all distribute stiffness the same way:
| Pattern | How it works | When to use it |
|---|---|---|
| Gyroid | Isotropic—stiff in every direction, no weak axis; slow to print because of its continuous curved path | Parts loaded from angles you don't control |
| Cubic | Stiff along all three axes, but with preferred directions; good stiffness per gram depending on load direction | Parts that take real load |
| Grid | Crosses in two directions within each layer; cheap and fast | Parts that mostly hold up their own top layers |
| Lines | A single direction per layer, alternating layer to layer; the fastest | Pure support infill, no structural load |
Don't lighten where stress concentrates
There's one place where lightening is exactly the opposite of what you want: stress concentrations. Stress doesn't spread evenly across the part; it piles up at abrupt changes in geometry. A sharp internal corner, the root of a rib, the edge of a hole, the start of an overhang: there the local stress can be several times the nominal stress in the rest of the part. Those zones aren't surplus; they're starved: they need more material, not less; a fillet that spreads the load flow, not a cutout that chokes it.
The classic mistake when lightweighting is to look at a part, see a zone that seems to do nothing, and hollow it out—when that zone was exactly where the load passed toward an anchor. That's why the eye isn't enough. Load paths aren't obvious, especially on a part with several supports or combined loads. Don't guess where material is surplus: measure it.
The rigorous way to know where there's surplus is the coupon: print the part, load it until it fails, and look at where it breaks. But read that break knowing the part is anisotropic: in FDM the crack usually starts along a layer line, in the weak Z direction, before it does where the nominal stress is highest. If it breaks along a layer joint, the fix may be to reorient the part, not to add material. Where it breaks for lack of section is where material was missing; where it survives intact is where there may be surplus.
Anisotropy matters as much as shape
Everything above reasons as if the part were a homogeneous solid—neutral axis, cube of the depth, load paths—and it's worth correcting that simplification before you cut anything. An FDM part isn't homogeneous: it's strong along the beads and weak between layers, and the print orientation decides which way that weakness points. It interacts with almost every technique in this article.
A rib printed upright stacks its layers in the very plane that must resist bending hardest—the weak plane; laid down, its beads run lengthwise and it resists far more. A thin shell inherits the between-layer weakness of any thin wall. And the test coupon misleads you if you don't separate "it broke for lack of material" from "it broke at the layer joint." So before you remove the first gram, decide how the part is oriented: sometimes the lightest and strongest version comes not from hollowing more, but from turning it so the layers run with the load. Layer adhesion and anisotropy and Orientation and overhangs develop this.
The loop: lighten, check, repeat
Lightweighting isn't a single cut; it's a loop. Start conservative. Remove material where the analysis—or the coupon—tells you that you can: hollow the center, swap infill for ribs, pocket the unloaded faces, open lattices where stress is low. Check again. Stop when the stress map spreads out evenly, with no large unloaded zones wasting material and no stress peaks with no material to take them.
A part designed this way weighs less, burns less filament, and prints faster than one you lightened by eye, and it stays just as strong exactly where it has to be. The stiffness was never in how much plastic you laid down, but in where you put it—and in which direction you stacked the layers. The next step is to fine-tune those reinforcements: Ribs, gussets and fillets tells you how to size the material you decide to keep so it works without becoming the weak point itself.