CAD, CNC, Modeling and Fabrication

We have been writing CAD software since it barely fit on the machines of the day. We started trying to make 2D and 3D CAD run on Commodore 64s and Amigas in the 1980s, and we have been at it through every generation of hardware since. We also do not just write the software: we run the machines. We are fabrication people who print, cut, and mill our own parts, so we understand CAD from the first sketch all the way to the chips on the shop floor.

Four decades of CAD

• In the 1980s we were writing 2D and 3D CAD code on Commodore 64 and Amiga hardware, back when fitting a modeler into a few hundred kilobytes was the whole problem.
• In 1989 we built custom sheet-metal duct software on a MicroVAX, the kind that lays out and unfolds ductwork into the flat patterns a shop can actually cut.
• In the early 1990s we worked on Progressive Peripherals and Software's 3D Professional and UltraDesign on the Amiga: 3D modeling and rendering and 2D drafting for a platform that was ahead of its time.
• We built MakeVR, a virtual-reality CAD and design tool, so you could model with your hands in 3D space instead of pushing a mouse across a flat desk. That work connects to our VR, AR, and xR practice.
• Today we work in SolidWorks and in OpenSCAD and the open parametric-modeling ecosystem, where the model is code and the geometry is driven by constraints.

Modeling, from freehand to parametric

We work across the ways geometry actually gets represented: boundary-representation solids, meshes, and parametric, constraint-driven models where a change to one dimension ripples through the whole part correctly. We move data between the formats the field really uses, including STEP, IGES, STL, DXF, OBJ, 3MF, and 3D PDF, and we repair the broken meshes and non-manifold geometry that real files are full of. That is the kind of work behind LGM's CADSpan, a tool we helped build that turns messy CAD into the clean, watertight STL a 3D printer needs. Sheet-metal unfolding and flat-pattern nesting, boolean operations, and the geometry processing underneath all of it are the same 3D graphics craft we bring to everything else.

Scan to CAD, and reverse engineering

Often the part exists before the model does. We turn real objects into editable geometry using computer vision: photogrammetry, structured-light and depth sensors, point clouds, and newer radiance-field methods like Gaussian splats and NeRFs that reconstruct a scene and hand it to a mesh. Getting from a noisy scan to clean, watertight, dimensionally honest CAD is a hard problem, and it is one we solve.

CAM, toolpaths, and G-Code

This is where a lot of software people stop and we keep going, because we understand the fabrication side by living on it. Additive processes like FDM, SLA, and SLS, and subtractive ones on CNC mills, lathes, routers, lasers, plasma, and waterjet. That means toolpath generation and CAM, slicers and post-processors, feeds and speeds, fixturing and work-holding, and the G-Code that actually drives the machine. When your CAD has to end up as a physical part, the pipeline from model to toolpath to finished piece is one we have walked end to end.

Visualizing and driving CAD

Big assemblies and digital twins have to be drawn fast and correctly. CAD visualization has been an OpenSceneGraph stronghold for decades, and we build the large-model rendering, inspection, and visual-simulation tie-ins that put a CAD model into a live, interactive context rather than a static viewport. We did extensive CAD simplification and optimization inside the open-source OSGWorks library, work funded by the Department of Energy through Iowa State University, so heavy assemblies stay fast enough to inspect and manipulate.

Who works on it

AlphaPixel is a US-owned small business, founded in 2004, with senior developers who have written CAD and modeling software since the 1980s and who fabricate their own parts, additive and subtractive. DLA DD2345 / ITAR registered. We take closed-source defense work and open source alike, and we do not sell a CAD package, so our advice is not a sales pitch for one.