First time trying to blend and mix various models. The goal was to create a model that would create realistic generic females hence the name, Generic Female Mix. Because of the nature of the models used to mix this, most outputs from this model will be NSFW in nature. Good prompt engineering will go a long way here.
The AUTOMATIC1111 Stable Diffusion Web UI Checkpoint Merger was used to create this merged model.
Recommended Parameters:
Steps: 15 minimum for a usable output. Results may vary.
CodeFormer: 0.79-1.0 is good. Increase for better face restoration. Can be disabled. Try to set to minimum 1.0 before opting to disable.
Width: 576 minimum.
Height: 576 minimum.
VAE: sd-vae-ft-mse-original. Can be disabled.
A resolution lower than 576x576 increases the chances for deformities and mutations. Lower at your peril.
Few prompts need more steps while a lot of prompts need fewer steps. Usually.
Don't leave the negative prompt empty. Make use of negative prompts. Even just one or two words would be fine.
Default output is NSFW so put 'NSFW' and 'uncensored' in negative prompt to suppress them and vice versa for more NSFW outputs. Put weights on 'NSFW' and 'uncensored' if needed for fine-tuning.
Description
Version 2.1 inpainting file.
Can do inpainting and outpainting.
FAQ
Comments (4)
its not working right with invoke.
What are the trigger words?
I have tried the exact same prompt and negative prompt from the first preview image with the checkpoint "generic femalemix_gefemiV21-inpainting.ckpt" in AUTOMATRIC1111 but I get totally different and much more worse result. I put the checkpoint file in \stable-diffusion-webui\models\Stable-diffusion\.Is there any thing else I have to do to get a result like in this image ? Thank you
If possible, can you add the smaller 2gb pruned+fp16 safetensor version ? I know that sometimes a model can't be compressed beyond a certain level.
Details
Files
Available On (1 platform)
Same model published on other platforms. May have additional downloads or version variants.