I created a new version using Krea 2.0 as a style LoRA. It’s still a draft, but there are already some decent results! Right now I need to generate some buzz, so I set a goal, but over the next few days I’ll expand the post with prompts and more info.
UPDATE 19/09: I hadn't noticed that the prompts for the various images were missing! I added the captions.
Description
Expanded dataset (77 images) and more detailed captions for each image. The lora currently includes 3+1 concepts: safe images, tentacles, shirikodama, and kotoribako (the latter, however, currently only has 2 images).