Stylize generations by adding LoRA models to Z-Image Turbo in ComfyUI

Example Style Test Prompt

Installing and Loading a LoRA in ComfyUI

1. Navigate to Civitai and download the desired LoRA model (e.g., Commodore 64 pixel art style).
2. Move the downloaded file into your ComfyUI installation folder under 'models' > 'loras'.
3. Open your ComfyUI interface.
4. Press the 'R' key on your keyboard to refresh the model list so the new file is recognized.
5. Double-click on the background canvas to open the node search.
6. Search for 'LoraLoaderModelOnly' (or 'Lora Loader Model Only') and select it to add the node.
7. In the new node, click the model dropdown menu and select the file you just downloaded.
8. Adjust the 'strength' value (e.g., set to 0.8 for 80% influence) to control how heavily the style impacts the generation.

Wiring the LoRA Node into the Workflow

1. Locate the main diffusion model node (Checkpoint Loader) and the sampling node (KSampler) in your current workflow.
2. Disconnect the direct link between the Diffusion Model and the KSampler (if one exists).
3. Click and drag from the 'Model' output of the Checkpoint/Diffusion node.
4. Connect it to the 'model' input slot on the 'LoRA Loader' node.
5. Click and drag from the 'model' output slot of the 'LoRA Loader' node.
6. Connect it to the 'model' input slot on the KSampler (or the next node in the chain).
7. Ensure the LoRA node sits legally 'in between' the source model and the sampler.

Chaining Multiple LoRA Models

1. Double-click the canvas and add another 'Lora Loader' node.
2. Select the second LoRA model from the dropdown list within the new node.
3. Disconnect the output from the first LoRA node.
4. Connect the 'model' output of the first LoRA node to the 'model' input of the second LoRA node.
5. Connect the 'model' output of the second LoRA node to the KSampler.
6. Repeat this 'daisy chain' process to stack as many models as required.