A proof of concept for Z-Image-Turbo Lora training.
Training Set: 12 images.
Training Steps: 3500 (v0.3), 3000 (v0.1 full), 1500 (v0.1 lite).
Recommended strength: 0.5 ~ 0.8 (v0.3), 0.7 ~ 1.0 (v0.1 full), 0.75 (v0.1 lite).
No trigger words needed.
Description
Retrained with the de-distilled model.