https://civitai.com/models/1612517/chinmi-and-tsuuhaiken-tekken-chinmi-and
Try size: 768 x 1024 or 960 x 1280 or 1024 x 1024 Weights: 0.8~1.0
Clip skip: 2Pictures from kantanmix_v16LCMTurbo
Sampling steps: 20~40 (It is suggestion 40 in duel / 雙人對決建議40)
Note: Note that the basic model I used to generate the image is LCM/Turbo, so the total number of steps and CFG are less. But in the same way, the two characters will inevitably make mistakes, and increasing the number of steps will reduce the error rate. It is easy to happen Tsuuhaiken. If you don’t like it, you can delete the trigger words, or strengthen the other posture description, or reduce the LoRA weight. / 注意我的生成圖片的基礎模型是LCM/Turbo，所以總步數跟CFG是比較少的。但同理，兩位角色難免會出錯，提高步數會使出錯率降低一些。很容易出現通背拳，假如你不喜歡可以刪除觸發詞，或加強其他姿勢描述，或降低LoRA權重。
Reference Video:
https://www.youtube.com/watch/qGMixvlopmw
