Usage ===== Step 1: Set Up ComfyUI Workflows -------------------------------- - Nunchaku workflows can be found at `example_workflows `_. To use them, copy the files to ``user/default/workflows`` in the ComfyUI root directory: .. code-block:: shell cd ComfyUI # Create the example_workflows directory if it doesn't exist mkdir -p user/default/example_workflows # Copy workflows cp custom_nodes/nunchaku_nodes/example_workflows/* user/default/example_workflows/ - Install any missing nodes (e.g., ``comfyui-inpainteasy``) by following `this tutorial `_. Step 2: Download Models ----------------------- First, follow `this tutorial `_ to download the necessary FLUX models into the appropriate directories. Alternatively, use the following commands: .. code-block:: shell hf download comfyanonymous/flux_text_encoders clip_l.safetensors --local-dir models/text_encoders hf download comfyanonymous/flux_text_encoders t5xxl_fp16.safetensors --local-dir models/text_encoders hf download black-forest-labs/FLUX.1-schnell ae.safetensors --local-dir models/vae Then, download the nunchaku models from `Hugging Face `_ or `ModelScope `_. For example, to download the Nunchaku Flux.1 dev models from Hugging Face, run: .. code-block:: shell hf download nunchaku-tech/nunchaku-flux.1-dev svdq-int4_r32-flux.1-dev.safetensors --local-dir models/unet/ .. note:: - For **Blackwell GPUs** (such as the RTX 50-series), please use our **FP4** models for hardware compatibility. - For all other GPUs, please use our **INT4** models. Step 3: Run ComfyUI ------------------- To start ComfyUI, navigate to its root directory and run ``python main.py``. If you are using `comfy-cli `_, simply run ``comfy launch``. Step 4: Select the Nunchaku Workflow ------------------------------------ Choose one of the Nunchaku workflows to get started. See :doc:`../workflows/toc` for more details.