AI/ML
Generate Videos Using WAN 2.1 on a Budget GPU - Full Setup Guide
Introduction
On February 25, 2025, Alibaba's Tongyi Lab launched Wan 2.1, an open-source video generation suite that can be accessed on Hugging Face and allows text-to-video production using consumer grade hardware. As of March 13, 2025, this paper explains how to install the Wan 2.1 T2V-1.3B model locally on a low cost GPU, like an NVIDIA RTX 4090, to produce a 5-second 480p video. It caters to developers and enthusiasts looking for easily available AI video tools in India or around the world.
Prerequisites
- Hardware: NVIDIA GPU with 8.19 GB+ VRAM (e.g., RTX 4090), 16GB RAM, 20GB free SSD space.
- Software: Ubuntu 20.04+ (Windows WSL2 viable), Python 3.8+, Git, CUDA 11.8+.
- Internet: Required for initial download.
Deployment Steps
Step 1: Install Dependencies
Update system and install essentials:
sudo apt update && sudo apt install -y python3-pip gitpip3 install torch torchvision --index-url https://download.pytorch.org/whl/cu118
Step 2: Download Wan 2.1 from Hugging Face
Clone the GitHub repo and download the 1.3B model:
git clone https://github.com/Wan-Video/Wan2.1.gitcd Wan2.1huggingface-cli download Wan-AI/Wan2.1-T2V-1.3B --local-dir ./models
Step 3: Install Requirements
Install dependencies from the repo:
pip3 install -r requirements.txt
Step 4: Generate a Video
Run the inference script with a text prompt:
python3 inference.py --model_path ./models/Wan2.1-T2V-1.3B --prompt "A cat jumping over a fence" --output video.mp4 --resolution 480p --duration 5Output: video.mp4 (5 seconds, 480p).
Requirements
Minimum System Requirements
- GPU: RTX 4090 (8.19 GB)
- RAM: 16GB
- Storage: 20GB SSD
- OS: Ubuntu 20.04+
Troubleshooting
- CUDA Errors: Verify CUDA version compatibility (11.8+).
- Memory Issues: Reduce batch size in inference.py if VRAM is insufficient.
Conclusion
Deploying Wan 2.1 on a local server with a budget GPU like the RTX 4090 is straightforward, leveraging its Hugging Face availability. This setup empowers Indian developers and hobbyists to generate 480p videos in minutes, showcasing accessible AI innovation. Wan 2.1’s lightweight 1.3B model makes video creation a reality without breaking the bank.
Ready to transform your business with our technology solutions? Contact Us today to Leverage Our AI/ML Expertise.
AI/ML