v2.0 - Trained on post-debut images using anylora + anything kl-f8-anime2 vae. amedira-mV model worked the best in my experience. Tends to be on the lewder side of things. SFW images (kinda broken) were uploaded.
v1.0 - Weights 0.6 - 1 with cfg 7 worked best. Teal colored hair worked better than blue hair for prompting. Civitai flagged every image submitted for review initially lol
