Ghost

Ghost - Generates only a close-up portraits for now, I don't know if I continue training for full height image. It seemed to me that it does not work very well with other LoRA.

Trained on SD 1.5
Triggers with Ghost, Ghostmask

To improve results recommended:
VAE sd-vae-ft-mse-original


Works well with ChilloutMix, NED,
Works randomly with 1.5 - sometimes good, sometimes the results are rubbish.


How to use LoRA's in auto1111:

• Update webui (use git pull like here or redownload it)

• Copy the file to stable-diffusion-webui/models/lora

• Select your LoRA like in this video

• Make sure to change the weight according to the instructions (by default it's :1)