What problem does it solve? Creating 3D assets for a Lens Studio Lens normally requires manual modeling or ad-hoc AI generation with inconsistent scale, orientation, and grounding. This Skill generates production-ready GLB meshes from text or image prompts, normalizes their size and facing, and reports AABB metadata so colliders and placement work correctly. ## Core Features & Use Cases - Four selectable backends: SPECS text-to-3D for high-fidelity static props, FAST3D for faster draft-quality generation, code-authored MeshBuilder for articulated or parametric TypeScript content, and Blender voxel for blocky or rigged GLB assets. - Asset-boundary normalization: Bakes target size, ground contact, and -Z forward facing directly into the GLB so downstream SceneObjects stay at unit scale and colliders match the visual. - Structured output contract: Every generation returns a report fragment with status, backend, AABB dimensions, center offset, orientation stamps, and completeness flags. - Use Case: Ask for a "red barn" prop for your Lens; the Skill picks SPECS text-to-3D, polls the async job, downloads the GLB to Assets/GeneratedMeshes/, normalizes it to your target size, and reports the AABB so you can place it foot-on-floor. ## Quick Start Ask the agent to build a 3D mesh, for example: generate a low-poly oak tree mesh about 40 cm tall for my Lens.
