AI artists continue to depend on AI image generators like Stable Diffusion, hence many asking how to use it with LoRA models. Basing art on LoRA is an easy way to get people to notice your art, as it can be based on actual people or fictional characters.
Granted, actual artists don’t need to use LoRA models to make works of art, but that’s not the topic of conversation. We do hope that AI art eventually leads to these artists making their own creations but that doesn’t seem likely right now.
For now, here’s how Stable Diffusion users can implement LoRA models in their artworks. How much work will this take from the artist? And will this cause any issues with their current style of making art? Read on.
Stable Diffusion - how to use LoRA models
Using Stable Diffusion to work on LoRA models is an essential technique for artists, as it can save a ton of storage space. While there are other ways to work on pieces of art, they can take up a lot of space, hence, the reliance on LoRA to smooth things out.
Whether you are accessing Stable Diffusion using Google Colab or running it locally on a PC, here is how to use LoRA models in Stable Diffusion:
- Make sure your AUTOMATIC1111 is fully updated and working, otherwise, this won’t work.
- Get your LoRA models from either Civitai.com or HuggingFace’s LoRA library.
- Once the LoRA is downloaded, put it on
stable-diffusion-webui/models/Lora. - On the SD checkpoint, save the LoRA model on
stable-diffusion-webui/models/Stable-diffusion - Fire up your WebUI and click Show Extra Network Icons under Generate
- Finally, click on the LoRA tab and simply choose which model you want to be added to your prompt
That’s how Stable Diffusion artists can use their LoRA models. It’s going to take some time and we’re sure some errors will be made but the efforts should be worth it. Once again, it would be better if AI artists actually drew things they liked but that’s neither here nor there.
In case you encounter the runtime error or the failed to load error while using the LoRA models, come back here as we have some easy fixes for these errors.
What are LoRA models?
In case any new Stable Diffusion users are unaware, LoRA models are basically files or images based on an existing person. This can be a model, an actor, or even an anime character, any of these can be considered LoRA models as long as the user has images of them.
LoRA is short for low-rank adaptation, which makes sense since these are just models used to make AI art. Granted, programming AI isn’t an easy thing whatsoever but any art that comes from it can feel like a joke.
Overall, it’s easy to see why AI artists rely on these LoRA models so they can recreate their favourite characters. We think actual drawings would be better but that’s just what AI art is like.
For more on Stable Diffusion, check out how you can generate NSFW images with the service. Users can also check out our Stable Diffusion Negative Prompt list to improve their art.