I spent an evening printing test brackets with the default cubic infill and a new geometry I had read about, then loaded both with weights from my kitchen scale. The new bracket held almost four times the load before it bent. Same filament, same printer, same wall settings. The only thing I changed was the internal pattern, and that change came out of a metallurgy paper I did not expect to care about. The result was the kind of thing that makes you rethink every part you have printed in the last few years.
Why my old brackets always flexed
When you slice a model in Cura, PrusaSlicer, or OrcaSlicer, the slicer fills the inside with a repeating pattern. Most users leave that pattern at the default, which is some variation of perpendicular struts arranged like a tiny egg carton. It is fast to compute. It is easy to print. It also handles load poorly, and the reason nobody flags that for you is because nobody running a hobby printer is testing structural stiffness under controlled conditions.
The UCL team behind the new pattern, working under Dr. Chu Lun Alex Leung and led by researcher David McArthur, set out to find a geometry that did a better job. They borrowed from metallurgy (the study of how metals are structured at the atomic level), where a phenomenon called crystal twinning has been studied for over a century. In twinned crystals, mirrored lattice planes carry load more cleanly than uniform ones. The team applied the same idea to 3D print infill, building a repeating cell with three extra mirror planes inside each unit.
What the new pattern does under load
Under load, every internal strut has two options. It can compress along its length, or it can bend sideways. Bending is what most cubic lattices do, because the right-angle junctions concentrate stress at the corners. Once a strut bends past its elastic limit (the maximum stress a material can take while still returning to its original shape), the whole structure folds.
The new pattern forces the struts into the second mode. When force is applied, the geometry steers the struts to align with the direction of load, which puts them in tension. Almost every printable plastic handles tension better than bending, because the polymer chains (long repeating molecules that make up the plastic) are aligned along the length of the strut. Pulling on the strut lines up with the direction the plastic is strongest.
The result, in the lab, was up to 380 percent more stiffness and 279 percent more strength before failure, using the same density of plastic. That is not a small number. That is a different part.
How I generated the geometry on my own printer
The pattern is not in any mainstream slicer yet, so the practical path is to generate the geometry yourself and import it as a mesh. A few realistic starting points:
- Download a triply twinned lattice generator from a research repository and feed it your part outline as a parameter.
- Model the cell in Fusion 360 or Blender, then use a Boolean operation (a 3D modeling step that combines, subtracts, or intersects two solid shapes) to fill the interior of your model with the new pattern.
- Search for community-published unit cell files on Printables or Thangs and import them as a starting point.
- Ask your local makerspace if anyone has a generator script running on a workstation.
Once you have a part with the new infill, print two test coupons (small standardized test pieces designed to measure a single property like stiffness) side by side. One with the new lattice, one with the default cubic lattice, both at the same density and same material. Load both with weights and watch how they fail. The numbers will not match the lab exactly. The trend should be obvious, and if your printer is calibrated well enough you should see a real difference even at small scale. If the trend goes the other way, the geometry may not have survived your slicer or your printer’s quirks, and the headline gains do not apply to your setup.
What this changes for the things you already print
For most of the prints I have on my shelves, the trade is the same. I want something light that does not flex under load. The new geometry changes that trade without asking me to change the filament. Less wobble in a camera mount, less vibration in a drone arm, less frustration when a hook snaps under the first load. The implications go beyond hobby prints. Aerospace and medical suppliers have spent decades chasing the same trade, mostly by switching to more expensive plastics or adding continuous fiber reinforcement. The UCL result suggests that geometry is doing more of the work than material choice, at least in the kinds of parts the team tested. The next time you reach for a more expensive filament to solve a stiffness problem, the cheaper fix might be in the slicer.
If the lab numbers hold up in real materials and real printers, the takeaway is bigger than one study. The internal structure of a printed part is a design variable, not a default setting.
Trade-offs
There are real reasons not to bet a critical part on this result just yet. The pattern is not in any mainstream slicer, so generating it takes modeling work and the tooling is rough. PLA, PETG, ABS, and resin each behave differently under stress, and the headline numbers came from specific test conditions. A hobby printer with stock firmware and a 0.4mm nozzle will not produce the same surface quality that the lab tested.
The orientation of the lattice inside your model also matters. The team found that rotating the part on the build plate can cut defect-related failures roughly in half, because most print defects show up on certain faces. Get the orientation wrong and the gains shrink fast.
Industry adoption will move slowly. Suppliers of certified parts will want their own data, repeated tests, and time on the calendar before they ship anything that depends on this. Expect the earliest practical use to show up in research labs, university engineering courses, and serious hobbyist communities first, with commercial design tools picking it up later.
What I would tell past me
Spend a few minutes on the geometry before you spend money on a fancier spool. Most of the wobble I have felt in my printed parts was a design variable, not a material limitation. The fix might be in the slicer settings you have been ignoring all along, or in a generator script you have not tried yet. The default infill is a starting point, not a verdict on what the part can do, and treating it like a fixed cost of 3D printing is the kind of thinking that makes you replace parts more often than you should. Try the new pattern on something cheap and disposable first, then judge for yourself whether the headline numbers hold up on your specific printer. That single test will tell you more about your setup than any benchmark review, and it costs you maybe twenty minutes of printer time.