Text/Image → Mesh → Rig

Build rigged 3D characters in one smooth pipeline.

Remesher is a practical ComfyUI-powered agent harness that takes a prompt or image and outputs a rigged GLB ready for animation, testing, and iteration.

View on GitHub ⭐ Star the project

Repository: https://github.com/remesher/remesher

What you get

End-to-end generation

Go from text prompt to rigged GLB in a single CLI command.

Image-edit support

Edit source character concepts and keep style/pose constraints.

Rig-ready outputs

Generate meshes and run rigging workflows for downstream animation.

How to use

Clone the repo:

git clone https://github.com/remesher/remesher.git
cd remesher

Install dependencies:
```
uv sync
```

Set ComfyUI server URL in config.json:

{
  "server_url": "http://localhost:8188/"
}

Command reference

Run these from the project root after uv sync.

Common flags

--out-dir <path>          # where outputs are saved
--mesh-seed <int>         # mesh generation seed
--texture-seed <int>      # texture generation seed
--target-face-num <int>   # target mesh face count

Health check

uv run comfy-prompt-cli health

Text → Image

uv run comfy-prompt-cli text-to-image \
  --prompt "A stylized game character concept"

Image + Prompt → Image (edit)

uv run comfy-prompt-cli image-text-to-image \
  --image path/to/input.png \
  --prompt "Add a large beard while keeping pose/style"

Image → GLB

uv run comfy-prompt-cli image-to-glb \
  --image path/to/input.png \
  --mesh-seed 12345 \
  --target-face-num 800000 \
  --texture-seed 67890

Rig GLB

uv run comfy-prompt-cli rig-glb \
  --mesh path/to/mesh.glb \
  --glb-name rigged_output

Text → Rigged GLB (end-to-end)

uv run comfy-prompt-cli text-to-rigged-glb \
  --prompt "A stylized game-ready character" \
  --out-dir downloads/smoke

Find it on GitHub

Source, issues, and updates live at github.com/remesher/remesher.

If this helps your workflow, click Star in the top-right of the repo page.