improved tagging to work better with human subjects
This LoRA is based on style of the artist Vasya Lozhkin.
LoRa trained on 203 images, edited out text.
Works well with turbo LoRa. at 1 cfg, 12-16 steps, as in the title images.
