Building Trellis 3D — turn images and text into initial 3D models

Turn images and text into initial 3D models for creative prototyping
No screenshots or demo video yet
Trellis 3D is an independent, browser-based tool for turning images or text descriptions into initial 3D models. Creators can rotate and inspect results in an interactive preview, then export assets in formats including GLB, OBJ, FBX and USDZ for further editing.
The workspace offers models such as Trellis 2, Hunyuan 3D, Tripo and Meshy, alongside a public gallery of generated examples. Background removal and image upscaling help prepare input visuals. It is designed for designers, game developers and creators exploring visual concepts or prototyping assets.
The product is built with Next.js and Codex-assisted development. It is a hosted independent service, not the official Microsoft TRELLIS research website. Generated models are a starting point for a creative workflow, not a promise of engineering precision or production-ready topology.
Everyone building this product. Owners can add team members and assign roles.
Building Trellis 3D — turn images and text into initial 3D models