Learning Timeline
Key Insights
LoRA Usage Limits
Avoid using too many LoRAs at once. It is recommended to limit usage to 2 or 3 LoRAs to prevent image quality degradation or excessive noise.
Advantages of Open Source Fine-tuning
Since Qwen Image Edit is open source, the community is constantly creating new LoRA files for specialized purposes that the base model cannot achieve on its own.
Step by Step
How to Add a LoRA to the Qwen Image Edit Workflow
- Right-click on an empty space on the ComfyUI canvas.
- Select the 'Add Node' menu or use the search function by typing 'Lora loader'.
- Select the node named 'Lora loader model only' from the search results.
- Disconnect the existing link connecting the 'GGUF' or 'Diffusion Model' to the next node.
- Connect the 'MODEL' output from the GGUF/Diffusion Model node to the 'model' input on the 'Lora loader model only' node.
- Connect the 'MODEL' output from the 'Lora loader model only' node to the 'model' input on the 'Ksampler' or the next processing node.
- Select your specific LoRA file (e.g., specialized/uncensored fine-tune) from the dropdown menu within that node.
- Repeat steps 2 to 6 if you want to chain multiple LoRAs together.
