What problem does it solve? AI-generated vehicles are the hardest 3D category: flat panels, sharp edges, and symmetric bodies expose every mesher flaw, from smoothing artifacts to melted rear ends. This Skill raises the success rate by enforcing multi-angle reference images, picking the right generation backend, and adding an optional detail-texture pass to recover panel crispness. ## Core Features & Use Cases - Multi-angle reference generation: Pass up to four view images via the imageUrls array so the mesher produces consistent front, side, and rear geometry instead of a melted back half. - Backend and polycount selection: Choose between hunyuan for curved vehicles and trellis for hard-edged designs, with polycount targets from 1.5k tris for distant scenery up to 60k for hero showpieces. - Scene wiring: Instantiate the generated .glb as a VehicleBody3D with wheels and collision for player cars, or as a lightweight MeshInstance3D for background traffic. - Use Case: A user asks for a player race car for a racing game; the Skill generates a multi-view reference, produces a 25k-tri mesh, imports it to res://assets/vehicles/, and wires it as a VehicleBody3D with a box collider. ## Quick Start Ask your agent to generate a sleek red Formula 1 race car with front and rear three-quarter views and wire it as a drivable VehicleBody3D in the main scene.
