Trained on manga images. LoRa supports three costumes: <ol><li>Elegant dress. | prompt: Elegant dress, cleavage</li><li>Swimsuit. | prompt: swimsuit, thigh strap, thighs</li><li>Pilot suit.| prompt: pilot suit, body suit, cleavage cutout</li></ol>For Version 2: increased the tagging quality with everything new i have learned. please note that some of the trigger words have changed.works well with str = 0.7-0.9if you get black and white colors add "monochrome, greyscale" to negative (see example images).

Files with SHA256: dfbb95ba54b9b7659751dbeec7c5387e927da370787883425c841a806244861a