RESEARCH & DEVELOPMENT
From game mesh to print-ready model.
We're building the technology to automatically convert raw triangular meshes into clean, printable geometry, no manual work required. Here's where we are.
What we're trying to do
We have 600+ car meshes extracted from video games. These are based on 1:1 real car scans, accurate panel gaps, badges, body lines, everything. But they're triangle meshes optimised for real-time rendering, not for 3D printing.
Our goal is to build an automated pipeline that takes any of these raw meshes and outputs a clean, watertight, properly segmented model that can go straight to a slicer. No manual cleanup. No retopology by hand. Just input a game mesh, get a printable car.
If we crack this, every car that's ever been scanned for a game becomes instantly available as a print-ready scale model. That's the vision. For now; once we get the 3D pipeline sorted and the base ready, then we're gonna let the AIs touch this goldmine... IYKYK
The problem with the current meshes
Game meshes are built to look good on screen, not to be physically manufactured. Here's what makes them unprintable out of the box:
Triangle-only topology
All quads are triangulated. This makes subdivision or smooth boolean operations nearly impossible without retopology.
The surface is millions of tiny triangles instead of clean rectangular patches. Like a mosaic vs smooth tiles. You can't reshape a mosaic without rebuilding it piece by piece.
Zero wall thickness
Game meshes are infinitely thin surfaces. Printing requires actual volume. Shells need thickness, parts need to be solid or properly hollowed.
The car body in-game is like a soap bubble, infinitely thin. To print something real it needs material thickness. Like the difference between a balloon and a plastic container.
We need your help
This isn't a closed lab. If you have experience in any of the following, we'd love to work with you:
3D Topology & Retopology
Manual or automated quad workflows
Computational Geometry
Mesh processing, repair, booleans
Machine Learning
3D segmentation, shape understanding
Python / Blender Scripting
Automating mesh operations at scale
3D Printing Experience
Testing prints, slicing, tolerances
CAD / Sub-D Modelling
Converting meshes to clean surfaces
Whether you want to contribute code, convert a few meshes manually, or just discuss approaches, all of it moves us forward.
Come hang out, ask questions, share what you're working on.
No gatekeeping, no "prove yourself first" nonsense. If you're curious, you're welcome.
Join the Discord, we don't biteWhat's next
Once one of our conversion approaches proves reliable at scale, we'll be expanding this page with full technical articles, documented pipelines, and before/after comparisons for every model. Until then, this is the working frontier.