Trained both Unet and text encoders (v1 only had Unet trained)
LoRa responds to prompts better now
Trigger for frilled green bikini works more consistently
Added character name as trigger word
Upped network dim/alpha from 16/16 to 32/32
Added photos of a similar-styled bikini to the images used to train to allow for more flexibility when prompting for the frilled green bikini
Trained both Unet and text encoders (v1 only had Unet trained)
LoRa responds to prompts better now
Trigger for frilled green bikini works more consistently
Added character name as trigger word
Upped network dim/alpha from 16/16 to 32/32
Added photos of a similar-styled bikini to the images used to train to allow for more flexibility when prompting for the frilled green bikini