Comment by @larryscruff • Hey

I really like these. What's a Lora though?

Stats

Comments

loras are like smaller custom models that are used alongside the underlying larger stable diffusion (usually) checkpoint models, you can kind of think of them like additional hyper fine tuning of a style that allows you to have greater precisionism for the output. so for example with the above synths, for the base model i used a model called mix-pro-v3-non-ema, and then i trained a custom Lora that was used in combination with this and more focused on pixel art x landscape styling , this probably explains it better: https://stable-diffusion-art.com/lora/