If that doesn't work, please try with CausVid lora.
CausVid lora strength: 0.3-0.5
Steps: 6-8
Cfg: 1.0
Video combine (trim of audio): true
(Option): cfg Zero
Lora model of CausVid:
https://huggingface.co/Kijai/WanVideo_comfy/blob/main/Wan21_CausVid_14B_T2V_lora_rank32.safetensors