What problem does it solve?
Rebuilding a real 3D object from reference photos often produces models that only look right from one camera angle. This Skill enforces an evidence-driven workflow that infers the complete parametric object — including hidden sides, scale, and manufacturing posture — before matching any reference view.
Core Features & Use Cases
- Real Object Brief: A mandatory written gate covering artifact identity, assumed scale, part boundaries, and inferred hidden-side geometry before any modeling begins.
- Camera Calibration & Comparison Boards: Calibrates one camera per reference image and renders side-by-side reference-vs-model boards with the bundled Pillow-based
compare_images.py script.
- Multi-Image Constraint Solving: Weights every reference image as a constraint on one shared geometry, iterating from blockout to canonical views to final inspection.
- Use Case: Given front, side, and rear photos of a mechanical bracket, produce a parametric
.forge.js model that holds up from all canonical views, passes forgecad run, and ships with comparison boards and inspection evidence.
Quick Start
Use the forgecad-reconstruct-from-images skill to infer the real object from these reference images and build a ForgeCAD model that holds up from all views.
