# Combining multiple LoRAs [[Low Rank Adapter (LoRA)]] are cool. What is even cooler is being able to combine multiple ones to generate images leveraging different models in a single generated image. This approach can be used to generate images combining different styles, different people, etc. ## How to combine different LoRAs? Justine Moore has created a short tutorial explaining how to do it with [[FLUX.1]]: - Choose the different LoRAs you want to use. Those should be public models (e.g., on [[HuggingFace]]) - Go to https://replicate.com/lucataco/flux-dev-multi-lora - Enter the prompt, and settings you want to use - In the `hf_loras` field, add the [[HuggingFace]] path to the models you want to use, or URLs pointing to the LoRA weights - Run Enjoy 🎉 ## References - Multi-LoRa image generator model: https://replicate.com/lucataco/flux-dev-multi-lora - Justine Moore's tutorial: https://x.com/venturetwins/status/1825924552817520896 - Related question on Reddit: https://www.reddit.com/r/StableDiffusion/comments/18dhvep/can_you_use_several_loras_simultaneously_how_do/ ## Related - [[Low Rank Adapter (LoRA)]] - [[AI image generation models]] - [[How to train a FLUX.1 LoRA]]