|
|
--- |
|
|
library_name: diffusers |
|
|
--- |
|
|
|
|
|
|
|
|
# MISHANM/video_generation |
|
|
|
|
|
The MISHANM/video_generation model is a diffusion-based video generation model . It is designed to generate high-quality videos from textual prompts using advanced diffusion techniques. |
|
|
|
|
|
## Model Details |
|
|
1. Language: English |
|
|
2. Tasks: Video Generation |
|
|
|
|
|
### Model Example output |
|
|
|
|
|
This is the model inference output: |
|
|
|
|
|
|
|
|
<video controls autoplay src="https://cdn-uploads.huggingface.co/production/uploads/66851b2c4461866b07738832/833kmbAdDGSxirLft0el7.mp4"></video> |
|
|
|
|
|
|
|
|
## How to Get Started with the Model |
|
|
|
|
|
## Diffusers |
|
|
|
|
|
```shell |
|
|
pip install git+https://github.com/huggingface/diffusers.git |
|
|
``` |
|
|
Use the code below to get started with the model. |
|
|
|
|
|
```python |
|
|
import imageio |
|
|
import imageio_ffmpeg |
|
|
import torch |
|
|
from diffusers import MochiPipeline |
|
|
from diffusers.utils import export_to_video |
|
|
|
|
|
# Load the pre-trained video generation model |
|
|
model = MochiPipeline.from_pretrained( |
|
|
"MISHANM/video_generation", |
|
|
# variant="bf16", |
|
|
torch_dtype=torch.bfloat16, |
|
|
device_map="balanced" |
|
|
) |
|
|
|
|
|
# Enable memory savings by tiling the VAE |
|
|
model.enable_vae_tiling() |
|
|
|
|
|
# Define the prompt and number of frames |
|
|
prompt = "A cow drinking water on the surface of Mars." |
|
|
num_frames = 20 |
|
|
|
|
|
frames = model(prompt, num_frames=num_frames).frames[0] |
|
|
|
|
|
export_to_video(frames, "video.mp4", fps=30) |
|
|
|
|
|
print("Video generation complete. Saved as 'video.mp4'.") |
|
|
|
|
|
|
|
|
``` |
|
|
|
|
|
## Uses |
|
|
|
|
|
### Direct Use |
|
|
|
|
|
The model is intended for generating videos from textual descriptions. It can be used in creative applications, content generation, and artistic exploration. |
|
|
|
|
|
### Out-of-Scope Use |
|
|
|
|
|
The model is not suitable for generating videos with explicit or harmful content. It may not perform well with highly abstract or nonsensical prompts. |
|
|
|
|
|
## Bias, Risks, and Limitations |
|
|
|
|
|
The model may reflect biases present in the training data. It may generate stereotypical or biased videos based on the input prompts. |
|
|
|
|
|
### Recommendations |
|
|
|
|
|
Users should be aware of potential biases and limitations. It is recommended to review generated content for appropriateness and accuracy. |
|
|
|
|
|
## Citation Information |
|
|
``` |
|
|
@misc{MISHANM/video_generation, |
|
|
author = {Mishan Maurya}, |
|
|
title = {Introducing Video Generation model}, |
|
|
year = {2025}, |
|
|
publisher = {Hugging Face}, |
|
|
journal = {Hugging Face repository}, |
|
|
|
|
|
} |
|
|
``` |
