ComfyUI-3D-Pack

Why we included this project

ComfyUI-3D-Pack makes 3D generation inside ComfyUI feel as turnkey as the image and video workflows already do. It accepts mesh and UV texture inputs and wires in a broad set of current methods, from Gaussian Splatting and NeRF to image-to-3D models like InstantMesh, TripoSR, TRELLIS, and Hunyuan3D, so you can go from a single reference photo or a rough scribble to a textured mesh without hand-coding each pipeline. If you already run ComfyUI for games, VFX, or product design, the prebuilt installers, Docker instructions, and ComfyUI-Manager support take most of the pain out of the underlying C++ and CUDA dependencies. Because so many methods sit behind one workflow interface, the suite also works well as a sandbox for comparing how different single-image-to-3D and splatting pipelines behave on the same input. Plan on installing system build tools and letting the larger models download before the heavier nodes run.