An attempt at Kanna. Trained exclusively on her 2 look since 3 lacked trainable content.
V2 patches the holes of V1, including a logic error where Kanna's face would turn green, now remedied with the tag "object head", as her head's a flower pot. How nobody thought of this before, I'm not too sure.
There's also "fishnet top" for a more stable, consistent look on the tied strings above her cleavage, and "black eyes, no sclera" has been reduced to "dot eyes" to hopefully save you bunch from a very scary encounter using a realism checkpoint. God...