Snaprint

SnapPrint (VoiceCAD) is an end-to-end multimodal AI pipeline that transforms a simple smartphone photograph and a spoken voice prompt into a 3D-printable CAD file (.stl). Designing custom functional 3D accessories (such as desk docks, wall mounts, sim-racing wheel adapters, or replacement brackets) traditionally requires expensive calipers, complex CAD software, and parametric modeling expertise. SnapPrint eliminates these barriers: Visual Calibration & Segmentation: The user snaps a photo of an object alongside a standard €1 reference coin. The system segments both the target object and the reference coin, converting pixel contours into sub-millimeter real-world dimensions via OpenCV calibration. Voice-to-CAD Generation: The user speaks their design intent (e.g., "Design an angled desk dock with a bottom cable cutout"). The system transcribes the command, extracts spatial constraints, and generates clean, modular OpenSCAD (.scad) code by leveraging a parametric CAD primitive library (lib_primitives.scad). Autonomous Validation & 3D Preview: An automated OpenSCAD compilation loop inspects the output, iteratively self-correcting syntax or geometry errors. The resulting .stl mesh is immediately rendered interactively in a Next.js / Three.js browser viewer ready for 3D printing.