HunyuanVIdeo in diffusers use negative_prompt but generate wrong video

[philipwan]

Describe the bug

Diffusers support negative_prompt for hunyuan_video recently, but when I use negative_prompt and set guidance_scale and true_cfg_scale, I got a video with all black elements. Maybe I set wrong parameters or save video fail.
How can I fix my problem? Thanks

Reproduction

import torch
import time
from diffusers import HunyuanVideoPipeline, HunyuanVideoTransformer3DModel, AutoencoderKLHunyuanVideo
from diffusers.utils import export_to_video, load_image, load_video
NEGATIVE_PROMPT = "Aerial view, aerial view, overexposed, low quality, deformation, a poor composition, bad hands, bad teeth, bad eyes, bad limbs, distortion"
model_path = "/realpath/hunyuanvideo-community-HunyuanVideo"
pipe = HunyuanVideoPipeline.from_pretrained(model_path, torch_dtype=torch.float16)
pipe.vae.enable_tiling()
pipe.to("cuda")
output = pipe(
prompt="The video shows a man and a woman standing in the snow, wearing winter clothing and holding cups of coffee. ",
negative_prompt=NEGATIVE_PROMPT,
height=480,
width=720,
num_frames=129,
num_inference_steps=10,
true_cfg_scale=6.0,
guidance_scale=1.0,
).frames[0]
export_to_video(output, "diffusers_480p_output.mp4", fps=24)

Logs

System Info

H20
resolution = 480 * 720
steps=10

Who can help?

No response

[github-actions]

This issue has been automatically marked as stale because it has not had recent activity. If you think this still needs to be addressed please comment on this thread.

Please note that issues that do not follow the contributing guidelines are likely to be ignored.

[haochengxi]

Maybe you should set true_cfg_scale=1.0, guidance_scale=6.0 instead?