Seo Sara

AI for AI: This LoRA was fully refined and brought to life using synthetic dataset engineering!

Since the original game lacks enough varied high-quality screenshots for proper training without severe pose overfitting, most of the dataset for this model was manually generated using Imagen / GPT. This allowed us to break the pose lock and train a truly flexible model across dozens of angles!

Because of the synthetic approach, the art style might slightly differ from the raw in-game render, but the overall accuracy and flexibility are miles ahead.

👗 Outfits Included:

1. Beachwear / Bikini Set: Her signature black criss-cross halter bikini, sheer blue shawl, and leg straps.

2. Volleyball Uniform (Setter S-Rank Version): Her canonical S-Rank blue/white setter jersey #2, shorts, knee pads, and optional fluffy jacket.

📌 Model Notes & Recommendations:

• Current Version: Epoch 8 (Released as the primary model because it offers the highest prompt adaptability, best pose dynamics, and lowest chance of breaking).

• Epoch 9–10: If you need higher outfit precision or stiffer anime lines, let me know in the comments—I can upload Epoch 8 or 9 upon request!

• Future Plans (V2): I'm already working on expanding the dataset even further to pull the style 100% closer to the original game render. Stay tuned!

edit: 8th epoch