Update v2: Updated my captions and lora training methods.Use cg for a cleaner look or stylize for a more artsy feel, combining the two work great too.Using realistic adds a lot of detail, but may warp (and sometimes fix) anatomy.Use traditional media for a marker/drawing style.Version 1:Works best with the anything-v4.5 modelWeight around 0.6 works best for me, anything below 1 seems to workwas trained to work with clip skip 2, but works well with skip 1

Artgerm style Lora

v1 was the first one I trained, it seems to need more details to have supergirl, like bodysuit, red cape, etc. May give better results in some cases.

Files with SHA256: 3ee0ff6bcc4bc356835e0fc5ed13889af825088cb40d0b5a63a5e18d004f4fcf