Tag
This paper presents a validated VLM-judge protocol for evaluating single-image-to-3D mesh quality, showing that cheap proxies like render-CLIP and geometry statistics fail to reliably track perceived quality.
UniMesh introduces a single model that jointly handles 3D mesh generation and understanding via a Mesh Head, Chain-of-Mesh iterative editing, and a self-reflection error-correction mechanism.
Modly is a local, open-source desktop app that turns photos into 3D models using AI, running entirely on GPU for Windows, Linux, and Apple Silicon macOS. It supports extensible models like Hunyuan3D and TripoSG, with workflow and CLI capabilities.