Halkali merkez neighborhood fatih st ozgur apt no 46 , Kucukcekmece , Istanbul , 34303 , TR
Discover the Wan2.2 series, one of the most powerful open source video models on Hugging Face. Experience this groundbreaking technology in text-to-video (T2V) and image-to-video (I2V) production with Eka Sunucu's high-performance GPU infrastructure.
Wan-AI is a powerful model family with Mixture-of-Experts (MoE) architecture, optimized for video production and editing tasks.
Text+Image to Video. The most suitable model for consumer electronics.
Text to Video. It produces cinematic videos just from text.
Image to Video. Animates images professionally.
Audio to Video. Lip sync and animation from audio.
Video creation processes with AI require much more computing power and VRAM than text-based models (LLM). Choosing the right hardware for Wan-AI models is critical to the success of your project.
| Model | Minimum VRAM | Recommended Card | Durum |
|---|---|---|---|
| Wan2.2-TI2V-5B Text+Image -> Video |
24 GB | RTX 4090 / 2x3090 | Home/Office Suitable |
| Wan2.2-T2V-A14B Text -> Video |
80 GB | NVIDIA A100 80GB | Server Required |
| Wan2.2-I2V-A14B Image -> Video |
80 GB | NVIDIA A100/H100 | Server Required |
| Wan2.2-Animate Video Editing |
48 GB | RTX 6000 Ada / 2x4090 | workstation |
Follow the steps below to run this model on an Eka Sunucu GPU server equipped with RTX 4090 (24GB VRAM) hardware or on your own computer.
It is recommended to use Ubuntu 22.04 LTS. Provides up to 30% performance increase compared to Windows.
Turkey (Turkish)
Germany (German)
Worldwide (English)
sudo apt update && sudo apt install python3.10 python3.10-venv -y
python3.10 -m venv wan_env
source wan_env/bin/activate
pip install --upgrade pip
Install PyTorch and required AI libraries. CUDA 12.1 version is recommended.
pip install torch torchvision torchaudio --index-url https://download.pytorch.org/whl/cu121
pip install diffusers transformers accelerate einops opencv-python safetensors xformers
Clone the model via Hugging Face. (Must have Git LFS installed)
git lfs install
git clone https://huggingface.co/Wan-AI/Wan2.2-TI2V-5B
cd Wan2.2-TI2V-5B
You can create a video at 720p resolution and 24fps with the following command.
python inference.py \
--prompt "cinematic drone shot of a futuristic city at sunset, neon lights, 8k, highly detailed" \
--image input_image.jpg \
--height 720 --width 1280 \
--frames 48 \
--fps 24
export PYTORCH_CUDA_ALLOC_CONF=expandable_segments:True You can optimize memory management by running the command. To get the best video from artificial intelligence, you must use the right words.
"Epic cinematic scene, golden hour lighting, slow motion, dramatic shadows, ultra detailed, film grain, anamorphic lens"
"Cinematic product video, soft studio lighting, slow camera movement, shallow depth of field, clean background, 4k look"
"A realistic human walking forward, natural body motion, smooth animation, cinematic lighting, stable camera, detailed face"
Negative Prompt Suggestion (What Don't We Want?):
"cartoon, anime, low quality, blurry, distorted, shaky, deformed, ugly, text, watermark"
Eka Sunucu's high performance Servers with Graphics Cards (GPU) Run Wan-AI, Stable Diffusion and other AI models seamlessly with . Installation support is from us!