Meet Jianna!
She was generated using Nano Banana (by Ivy, thank you Ivy!!!) and Qwen-Image-Edit for style and outfit transfers, to create the dataset!
Description
This works on base or distilled.
Trained on images with a 1024 resolution. It gets a little unpredictable if you use a latent above 1024, so upscale beyond that if you want to keep consistency.