LoRA of TV-host Stacey Dooley.

This will be the last version of this LoRA. Somewhere between doing some test generating with a realistic model and Anything V3-VAE (and not realizing why it looks like garbage) I accidentally deleted the whole training folder. Still had the LoRA-file and some training information though.

V2
Trained on 127 images (new/corrected dataset, 768x768, 3 epochs, 2322 steps). Trigger should not be needed if <lora:St4c3yD00l3y_V2:x> (x = weight) is included in the prompt.

Useful prompts: brown hair, long hair, blonde hair, blue eyes, orange hair, smile, teeth, grin, open mouth, earrings, hoop earrings, necklace

V1
Trained on 126 images (768x768, 3 epochs, 2079 steps). Trigger should not be needed if <lora:St4c3yD00l3y_V1:x> (x = weight) is included in the prompt.

Useful prompts: smile, teeth, jewelry, long hair, looking at viewer, earrings, light ginger hair, brown hair, blonde hair

It is recommended to use the V1 LoRA together with the Stacey Dooley V1-embedding. This will help in achieving better likeness and more consistent results than just using the LoRA or the embedding separately. The LoRA weight should be around 0.5 when combining them. Example to include in prompt: St4c3yD00l3y_V1-emb, <lora:St4c3yD00l3y_V1:0.5>