Custom SDXL LoRA

Upload at least 20 images to create a new LoRA model that can be used to generate your own custom subject or style.

What is a Custom LoRA?

Low-Rank Adaptation (LoRA) is reshaping machine learning technology by providing a way to quickly fine-tune Stable Diffusion models, allowing you to create just about anything with the right training data. By focusing the training process on specific concepts, from characters to specific styles, LoRAs can be a quick and lightweight way to add new data to Stable Diffusion and create incredible custom artwork.

What is SDXL?

SDXL 1.0 is a Stable Diffusion model that produces results at a higher resolution than previous models. The model can output images natively at resolutions around 1024×1024, while previous models could only produce images around 512×512.

SDXL LoRAs should be added to the base SXDL model for ~90% of the steps, allowing the refiner to complete the image without the LoRA.

Create a Custom LoRA:


    This site is for sharing ai generated images and other media along with the techniques to recreate them. AI generation is a process that benefits greatly from experimentation, iteration and knowledge sharing. This is why our goal is to help people create and share incredible art. Let’s make something beautiful!