stable-diffusion text-to-image

.safetensor model for automatic1111 webui.

Strange_Dedication_v3 is an improvement to Strange_Dedication_v2 using Anything_v4.5. It's better at the cutesexyrobutts style, without having to use a trigger. Also, it's good at shiny_skin and shiny_clothes and artistical backgrounds.

I have only used it with "vae-ft-mse-840000-ema-pruned", CLIP-Skip 1 and with danbooru tags. Lately I have started using the negative embed "bad-hands-5" (by an unknown author?), which was used for the example images as well. If you work with those you should be able to prompt images like these (prompts in .png metadata):

00009-20230122002457-6cda57b672.png 00009-20230121201847-6cda57b672.png 00007-20230122001758-6cda57b672.png 00002-20230122000635-6cda57b672.png