"Please use it together with both the High-Noise and Low-Noise versions."

My previous character LoRA tended to produce eyes that looked overly enlarged, similar to app-edited results. To address this, I iterated and fine-tuned the model, and this release aims for a more natural, everyday look.

Please note that results may vary depending on the companion LoRAs, prompts, and seed you use. As a result, the face in your outputs may differ from the sample video.

The workflow, companion LoRAs, and prompts used in the sample video are documented for your reference.