A textual inversion of the actress Emma Stone. Enjoy!
Parameters for success
CFG 5-6
No LATENT Hires fix, use 4x Ultrasharp or another non-latent instead. (Denoising around 0.2 or 0.25.)
Add the word "young" before "3mmaston3" to give her a more put-together appearance.
Specify certain hairstyles, as it defaults to slightly shorter hair.
Works best on photorealistic models, but can be used with illustrated models too.
Works great at 512x768 and 768x1024.
Description
FAQ
Comments (1)
Great model thanks, Could you share some info how it's trained or a video to follow, I trained my face two times, using Lora and got the worst result, but haven't tried Textual inversion