I am trying to run stable video diffusion with the code below but got all black videos (which is composed of all black images). It seems to be caused by the fp16 format. After trying your VAE, I got this error: "RuntimeError: Input type (float) and bias type (c10::Half) should be the same". Any suggestions for how to fix?

#################################code###

import torch

from diffusers import StableVideoDiffusionPipeline, AutoencoderKL
from diffusers.utils import load_image, export_to_video

pipe = StableVideoDiffusionPipeline.from_pretrained(
"stabilityai/stable-video-diffusion-img2vid", torch_dtype=torch.float16, variant="fp16"
)

pipe.to("cuda")

Load the conditioning image

image = load_image("https://huggingface.co/datasets/huggingface/documentation-images/resolve/main/diffusers/svd/rocket.png")
image = image.resize((1024, 576))

generator = torch.manual_seed(42)

export_to_video(frames, "generated.mp4", fps=7)