Hinata v2
An upgraded version of PixAI's Hinata model trained on images up to 2K, with sharper details and more flexible compositions.
Use this model on PixAI
2,791 likes · 3,010,005 generations
Hinata v2 is an upgraded version of PixAI's Hinata model, trained on high-resolution images up to 2K for sharper detail and greater creative freedom. It can also be used as a base model for LoRA training.
Key Features
Improved Quality — Sharper details, including clearer hair and eye features.
Enhanced Aesthetic — Flexible and diverse compositions for even more creative freedom.
Simpler to Use — Simple tags continue to deliver beautiful results with ease.
Higher Resolution Training — Trained with high-resolution images up to 2K, allowing for more detailed outputs.
Tips
We recommend avoiding "quality" tags to prevent introducing unexpected biases in your results.
The Rescale CFG and Shift features are unlocked for both Hinata and Hinata v2, allowing more refined control over your image outputs:
Rescale CFG (0.0–0.7) — Adjust this to fine-tune image exposure and saturation. Higher values help correct overexposed or oversaturated images.
Shift (1.5–10) — Controls the level of fine detail in generated images. Higher values bring more detail but may reduce sharpness and vibrancy; lower values create cleaner, more varied results.
