When use Wan2.1-i2v-720p-14B to infer, the generate video is bad with color cast?

[haozheguo]

When I use the original weight of wan i2v-720p-14B, the generate result's color is incorrect and effect is bad. the infer config are as follows:
CUDA_VISIBLE_DEVICES=7 python wan_generate_video.py --fp8 \ --task i2v-14B --video_size 1280 720 \ --video_length 81 --infer_steps 20 \ --prompt "The scene features an animated girl with long, wavy hair adorned with blue and black hair accessories, wearing a black and white outfit with a choker. She is sitting in a room with a window in the background and a decorated Christmas tree to her right. The girl appears to be speaking, as her mouth moves slightly, and the camera gradually zooms in on her as she continues to speak." \ --save_path ./infer_wan_debug \ --output_type both \ --dit ./pretrained_weight/Wan_2.1_ComfyUI_repackaged/split_files/diffusion_models/wan2.1_i2v_720p_14B_bf16.safetensors \ --vae .pretrained_models/HuggingFace/Wan-AI/Wan2.1-I2V-14B-720P/Wan2.1_VAE.pth \ --t5 ./pretrained_models/HuggingFace/Wan-AI/Wan2.1-I2V-14B-720P/models_t5_umt5-xxl-enc-bf16.pth \ --clip ./pretrained_models/HuggingFace/Wan-AI/Wan2.1-I2V-14B-720P/models_clip_open-clip-xlm-roberta-large-vit-huge-14.pth \ --attn_mode torch \ --image_path ./images/8b1eb8cf56614c24bba841cd3168a517_3822.png
the generate result are as follows:

Uploading 20250610-192623_2133027379_.mp4...

[haozheguo]

the input image:

the random sample frame from the video:

[lsx193201]

Hello, may I ask if you have found a solution to this problem? I have also encountered the same issue and I can't figure out the cause.