Recommended weight: 0.8 to 1Just got a RTX 3090, so I could train the text encoder along with the Unet this time and also increase the network dim/alpha for v2.

Emma Verde [Love Live Nijigasaki] SDXL

<ul><li>Trained both Unet and text encoders (v1 only had Unet trained)<ul><li>LoRa responds to prompts better now</li><li>Trigger for frilled green bikini works more consistently</li></ul></li><li>Added character name as trigger word</li><li>Upped network dim/alpha from 16/16 to 32/32</li><li>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</li></ul>

Files with SHA256: 3a67f3f240718a3d0a96494e9cfe8686b52948ed924ce44061baf81443b81826