Made To:
Play with gender. Trained on ML generated images of single subjects with a wide variety of body types and complexions, with a focus on mixing masculine and feminine traits to produce new combinations.
Rather than specific trigger words, this one was trained on auto-generated booru tags, to see what a data set of different gender combinations would do with a fairly standard tagging format.
If you read the above as "gender-bendy mystery-box", you have the right idea.
Model Weights
Most of the gallery images were between 0.2 - 0.5, depending on what checkpoint and other loras were used.
Similar/Related LoRAs:
Enby/Andro Mix used a similar data set with more targeted keywords to give a more controlled effect.