スポンサーリンク
A mage2Video model called stable video diffusion, which generates video from images, is now available.
touch-sp.hatenablog.com
huggingface.co
I tried to use google colab to try stable video diffusion, but it stopped at the model loading point.
I checked how to deal with it.
This article is a reminder of how to solve the problem.
contents
スポンサーリンク
abstract
How to resolve hang-up in StableVideoDiffusionPipeline.from_pretrained
1.requirement
Google Colab
Diffusers == 0.25.0
2.issue
When I run the following sample code, I get a situation where I can't proceed from the "StableVideoDiffusionPipeline.from_pretrained" point
I got a situation where I could not proceed forever!
import torch
from diffusers import StableVideoDiffusionPipeline
from diffusers.utils import load_image, export_to_video
pipe = StableVideoDiffusionPipeline.from_pretrained(
"stabilityai/stable-video-diffusion-img2vid-xt", torch_dtype=torch.float16, variant="fp16"
)
pipe.enable_model_cpu_offload()
# 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)
frames = pipe(image, decode_chunk_size=8, generator=generator).frames[0]
export_to_video(frames, "generated.mp4", fps=7)
3. how to solve
They say you can work around it by running the following code
from huggingface_hub.utils import _runtime _runtime._is_google_colab = False
The improved code is as follows
import torch
from diffusers import StableVideoDiffusionPipeline
from diffusers.utils import load_image, export_to_video
from huggingface_hub.utils import _runtime
_runtime._is_google_colab = False
pipe = StableVideoDiffusionPipeline.from_pretrained(
"stabilityai/stable-video-diffusion-img2vid-xt", torch_dtype=torch.float16, variant="fp16"
)
pipe.enable_model_cpu_offload()
# 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)
frames = pipe(image, decode_chunk_size=8, generator=generator).frames[0]
export_to_video(frames, "generated.mp4", fps=7)
Successfully executed after modification.
スポンサーリンク
