follow me Paranoid American Instagram, Youtube, Patreon and the website.

ODB is for the children!

follow me on Instagram, Youtube and my website.

Suggested weight of 0.8-1.0 seems to work well. Realistic Vision 1.4 seemed to get the likeness better than RV1.3. Tried on a few others - aside from very close portraits the likeness gets lost easily and requires inpainting.

Trying to combine with certain subjects (like "Big baby Jesus" for example) will result in him having lighter skin so for those you'll want to turn the weight up.

Also seems to react better to higher CFG values for certain prompts, between 7-11.

This is still a WIP, but have to start somewhere. This was trained on only 6 low quality images with the following settings:

Learning Rate: 0.005:1000, 0.001:2000, 0.0001:5000, 0.00005

Batch Size: 1

Gradient Steps: 1