Latent Consistency Model (lcm) is a new model to improve the speed of image generation for Stable Diffusion. You only need 4 steps to generate a high quality image. I am going to show you how to install this on Automatic1111’s webui as an extensiton.
Enable the Extension
- Click on Extensions tab and then click on Install from URL.
- Enter https://github.com/0xbitches/sd-webui-lcm in the URL box and click on Install.
- Click on Installed and click on Apply and restart UI.
Text to Image Example
- After UI is restarted, you can see a new tab LCM. If the tab is not showing, please restart webui. Click on the LCM tab and then LCM txt2img
- Enter your prompt. Expand the Advanced option and change the setting as needed. Note that there is no negative prompt, and the default batch count is 4 and the default steps is 4.
- Click on Generate to begin generation. When you do it the first time, it downloads the models from the repo and it might take a long time. After you run it a few times, it should be fast. On my Nvidia RTX 2080Ti, it takes about 0.4 ~ 0.5 seconds for a 512×512 images and 4 steps. By default, the images are saved under outputs\txt2img-images\LCM .
When I tried to run img2img and video2video, I got this error.
TypeError: slice indices must be integers or None or have an __index__ method
I will update this post again when this error is fixed.
This post may contain affiliated links. When you click on the link and purchase a product, we receive a small commision to keep us running. Thanks.