Node to set v-prediction sampling when using SDXL and other models that may not have the necessary metadata to identify it as a v-prediction model. This node is useful for quantized models since they lack the necessary metadata.
Node to set v-prediction sampling when using SDXL and other models that may not have the necessary metadata to identify it as a v-prediction model. This node is useful for quantized models since they lack the necessary metadata.
The workflow is simple.
Add the "AddParam" node to your ComfyUI workflow.
Connect the "AddParam" node to the model loader node (e.g., the UNetLoader).
Select "epsilon" or "v_prediction" from the dropdown menu.
Use GIT:
Navigate to the custom_nodes
folder inside the ComfyUI directory.
cd ComfyUI/custom_nodes
Git-clone this repository
git clone https://github.com/magekinnarus/ComfyUI-V-Prediction-Node.git
Restart ComfyUI after installing, and the node will be available for you to use.
ComfyUI: this node utilizes the modules in the ComfyUI codebase for its functions.