This is a LoRA used to enhance running posture, which can automatically generate running movements for the front and back of the character based on the reference image.
Upload a workflow of automatic prompt words and translations that I use.
Recommend a LoRA weight of 0.3~1.0.
Setting a high LoRA weight will slow down the video speed, while setting a low LoRA weight will make the video speed approach normal playback speed.
>>> Wan-i2v Auto Promprt Translate-Kijai-480p
purgevram
Description
V1.0 is not satisfactory, skip it and release v2.0 version
FAQ
Comments (7)
我一点想不明白，wan号称各项数值都优异过其他视频模型，连个那么基本的人类行走和跑步的日常动作都要用lora来控制，这。。。。。。。无语。。。
除了可灵，其他的视频模型走路跑步都不太行
jtddwcj3814Apr 29, 2025
真要把那么多动作塞进模型里，你显存又不乐意了，
How to generate high-definition videos。如何生成高清晰度的视频？
如果你显卡给力的话，直接生成720P视频就很清晰了，不够的话还可以再ai放大到1080p。
Can you write the prompt ? I don't use comfi
Any chance you could make this for Wan 2.2 14B ??