Scanning small objects accurately is where cheap commercial hardware usually disappoints, and this build tackles exactly that gap. The maker has spent nearly a decade on accessible DIY scanners and argues that photogrammetry remains an undervalued method, delivering strong results for a fraction of what market scanners cost. The demonstration subject is suitably humbling: a black DVI-to-VGA adapter, captured with an OpenScan Classic and the newly released Macro Add-on.
The rig itself is refreshingly tangible. Blue printed frames and large plastic gears sit on a wooden base, with stepper motors placing the subject before a digital camera surrounded by an LED ring light. A cross-polarized ringlight removes shadows and reflections, and focus stacking runs on the device itself, giving what the maker describes as near-unlimited depth of field. Photos were taken from 200 positions and stacked, then processed in Metashape, RealityScan, or OpenScanCloud, with Blender handling the textured renders.
The numbers are the strongest part. The measured VGA hole distance came out at 2.279 mm with a spread of 0.030, and pin centers showed a standard deviation of 0.006 mm at 1-sigma, within manufacturing tolerances. The dense cloud passed 21 million points and the mesh holds 4 million triangles. Materials were segmented by hand; only the render workflow got AI help, with a language model suggesting materials and camera poses. That final render took over ten hours.
Two open points stand out. The DVI-side pins still need some kind of correction, as with nearly all scanners. And the rising Raspberry Pi price is a real issue, with alternatives being sought, while running the firmware from PC or Mac still needs a way to control the GPIOs. A scan wizard that determines camera settings automatically is apparently close to release.