You use 1 ~0.8.

Stabilization and dataset improvements.

The learning model is animefull-final-pruned.
Generation model is AbyssOrangeMix2_sfw.