Trellis MCP Server

The Trellis Blender Plugin brings Microsoft’s state‑of‑the‑art text‑to‑3D and image‑to‑3D generation model directly into the familiar Blender environment. By exposing TRELLIS’s powerful AIGC capabilities as a native addon, the server eliminates the need for external pipelines or command‑line tooling. Developers can now generate richly textured meshes from natural language prompts or reference images, then immediately manipulate and refine them within the same 3‑D workspace. This tight integration reduces iteration time, keeps creative assets in a single context, and opens the door for more complex downstream workflows such as procedural asset libraries or AI‑driven level design.

At its core, the plugin offers four generation modes: plain text‑to‑3D, image‑to‑3D, and two conditional refinement paths that combine an existing mesh with either text or image guidance. Each mode leverages TRELLIS’s diffusion architecture, allowing users to tweak sampling steps and classifier‑free guidance strengths for fine‑grained control over detail fidelity versus stylistic alignment. Post‑processing options—mesh simplification, texture resolution, and baking mode—provide additional polish without leaving Blender. The asynchronous request handling ensures that the UI remains responsive, while real‑time status updates and clear error messages keep users informed about connectivity or processing issues.

One of the standout features is the built‑in MCP integration. By launching an embedded MCP server, the addon can expose its capabilities to any MCP‑compliant AI assistant such as Claude, Cursor, or Windsurf. Developers can simply paste a concise configuration snippet into the assistant’s settings to register “trellis” as an available tool. This seamless bridge means that a conversational AI can now ask the Blender instance to generate or refine a mesh on demand, enabling new forms of interactive design sessions and automated content pipelines.

Real‑world scenarios abound: a game artist can request a stylized prop from a textual description, immediately preview it in the scene, and iterate; a VR designer can generate environment meshes from reference photographs on the fly; or an educational tool could let students experiment with AI‑generated geometry without leaving the 3D viewport. Because the plugin communicates over a local API, it can also be scripted or integrated into larger production pipelines, making it an attractive component for studios that need rapid prototyping alongside rigorous version control.

In summary, the Trellis Blender Plugin transforms a powerful research‑grade 3‑D generation model into an accessible, workflow‑friendly tool. By uniting text/image prompting, conditional refinement, post‑processing, and MCP connectivity within Blender, it empowers developers and artists to harness AI‑generated geometry directly in their creative environment—accelerating iteration, reducing context switching, and opening new possibilities for AI‑augmented 3‑D content creation.