English
English
Imagine you have a versatile robot chef who can make a variety of basic dishes. However, if you want this robot chef to make special, specific dishes, such as Italian pizza or Chinese dumplings, you need to give it some extra "recipes".
1.Stable Diffusion Stable Diffusion is like this versatile robot chef, which can make a variety of dishes (pictures) based on different ingredients (descriptions). It is very good at understanding your requirements and creating delicious dishes that fit your description.
2.LoRA Model The LoRA model is like an extra "recipe" provided to this robot chef. These “recipes” can help robot chefs learn to make special dishes to enrich their cooking skills.
1. Basic Ability: Stable Diffusion already knows how to make various basic dishes.
2. Special Recipes: LoRA is those special “recipes” that teach chefs how to make dishes with a specific flavor.
3. Learning process: By learning these "recipes", robot chefs can gradually master the skills of making these special dishes.
1. Personalization: Using LoRA, robot chefs can create more personalized dishes that meet your needs for specific flavors.
2. Flexibility: Even if you learn the new “recipes”, robot chefs can still make any other dish, which is like it becomes more versatile.
3. Efficiency: LoRA allows robotic chefs to learn new “recipes” without forgetting their original cooking skills so it can create the dishes you want faster.
The LoRA model in Stable Diffusion is like a special set of “recipes” for robot chefs, allowing it to create the dishes you want more flexibly and personal while maintaining its original cooking capabilities.
Most AI drawing communities have model downloads. Filter LoRA in the corresponding website filters to quickly find the model of this type.
civitai: https://civitai.com
liblib (domestic in China): https://www.liblib.art/
As shown in the figure below, select LoRA when civitai filtering models to filter out LoRA models civitai_filters
Store the downloaded model in the "ComfyUImodelsloras" directory, and then restart or refresh the ComfyUI interface to load the corresponding LoRA model.
Because the model needs to distinguish versions, in order to facilitate later use, it is recommended that you rename the model file and add a model version prefix such as "SD1.5-Model Name", or not rename it. Create a new folder in the corresponding model directory, name it with the big model version, and then copy your model file into it like "ComfyUImodelslorasSD1.5".
If you are sharing model files with other GUIs, please refer to the instructions on how to share the model part of ComfyUI and other Stable diffusion AI drawing WebUI, and copy and install the corresponding model file to the corresponding folder.