@annabel.lucinda on Instagram
- v2.0: Trained on a bigger dataset of 57 3:2 images of her. Also first attempt to include her body. The LoRA is heavily biased towards gym clothes, for obvious reasons.
- v1.0: Trained on 20 1:1 (768x768) images of her face only.
Use of any LoRA to enhance details is recommended for better results.