Put slightly more effort into tagging and added triggerwords to separate her outfit
Pruned dataset
Better cropping
Needs better tagging.
Could probably improve file size a fair bit.
Trained on AOM3 (d124fc18f0)
Lop from Star Wars Visions.
