WAS Hajime Sorayama
A Hajime Sorayama Textual Inversion Faux Style
Usage:
Ending your prompt with the following segments is probably the best method. Name may differ for different versions.
art by WAS-HSart by [(WAS-HS):0.5] (Less effect)art by (WAS-HS) (More Effect)art by [((WAS-HS)):1.5] (Extreme Ultra Mode)... etc
Example Positive Prompts:
A portrait of a android, art by WAS-HS
Notes:
Needs more data, but alas, I couldn't find it. So probably mixed with good prompts, and other styles mainly.
❤ Hearts and 🖼️ Reviews let me know you want moarr! :3
Description
Version 2 of Hajime Sorayama style, with a larger dataset (around 400 images total after clean-up and restoration) thanks to atomicnixon!
Still has a bias towards women due to the overwhelming ratio of women to men, but you can use weighting to get buy that. Now that it can do men much easier, it will be easier to garner more data for a more balance model.
If you would like to help with submitting men and male robot androids, that would be much appreciated ❤️ Preferably in 512x512 gens or cropped to.
FAQ
Comments (4)
FABULOUS ! AMAZING !
10/10 STARS !
Thank you so Much !
"Needs more data, but alas, I couldn't find it." - can you tell us about your training settings, or share the training data set with us? I'd like to help you develop this one, as it's a very cool style that I'm interested in.
Could you share an example prompt and a model you use? My outputs with this lora are awful and nothing like the great samples you posted.
C3P0?