If you love action lora, you can come and discuss via discord channel

可以通过 QQ 3877209711 联系我

An action lora for I2V. Hadouken is from Street Fighter series.

version 1.0, train on low resolution and 1200 steps

Notice: Only work on WAN2.1-I2V-720p model，maybe work on 480p-i2v as well

注意： 仅作用于WAN2.1-I2V-720p模型，480p的i2v模型可能也行

Sugguested strength: 1.0, suggested step: 30 (You can use Teacache to accelerate). Placing the main character on the left side of the image is suggested. Maybe 30 to 40 frames is enough for Hadouken.

建议强度1，建议步数30(最好用Teacache加速，不会有太大影响)。建议把角色放在图片的左边。我感觉30到40帧就足够完成这个动作了。

The trigger word is, someone is lowering one's body with legs wide apart and shooting a blue energy ball with two hands.

eg. Donald Trump is lowering his body with legs wide apart and shooting a blue energy ball with two hands.

Also, you can define the detail appearance of the main character before the trigger words.

触发词是：someone is lowering one's body with legs wide apart and shooting a blue energy ball with two hands.

例如：特朗普 is lowering his body with legs wide apart and shooting a blue energy ball with two hands.

你可以在触发词之前，详细描述该角色的外表特征

I strongly suggest you use the work flow of kijai. You can find my workflow screenshot on the post

https://github.com/kijai/ComfyUI-WanVideoWrapper

我建议你使用kijai的工作流。你可以在post找到我的工作流