I wanted to combine multiple emoticons into one file, but it was difficult to avoid the problem of them gradually mixing up. While this is a significant loss in both file size and production time, I think this is the most reliable method.

It's recommended to set the Lora weight between 0.8 and 1. If the shape is occasionally distorted, using a detailer on the inside of the speech bubble will correct it.

Be sure to include the < > around the trigger word. This can significantly reduce the impact of emoticon-related tags on the character's expression.