With the experimental Latent Consisteny Models, we can use like a default LORA, we can generate images way faster than with previous methods. There are a lot less steps and CFG scale involved in this – check out the infos over here: https://github.com/0xbitches/sd-webui-lcm
![](https://turboflip.de/wp-content/uploads/2023/11/griddy-scaled.jpg)