README.md

metadata

title: Wan 2.2 Text + Image to Video
emoji: 🔥
colorFrom: purple
colorTo: indigo
sdk: gradio
sdk_version: 6.1.0
app_file: app.py
pinned: true
license: apache-2.0
short_description: Text + Image to Video generator using Wan 2.2 5B model

Wan 2.2 Text + Image to Video Generator

Generate high-quality videos from text prompts and optional images using the Wan 2.2 TI2V-5B model.

Features

• Text-to-Video: Generate videos from text prompts only
• Image-to-Video: Animate static images with text prompts
• Full Parameter Control: Adjust all generation parameters including:
  • Sampling steps (10-50)
  • Guidance scale (1.0-10.0)
  • Sample shift (1.0-20.0)
  • Solver selection (UniPC or DPM++)
  • Custom resolution (multiples of 32)
  • Duration control (0.3-5.0 seconds)
  • Negative prompts
  • Seed control for reproducibility
• ZeroGPU Support: Optimized for Hugging Face Spaces with ZeroGPU hardware

Model

• Model: Wan-AI/Wan2.2-TI2V-5B
• Paper: Wan 2.2 Paper

Usage

1. Text-to-Video: Leave the image input blank and provide a text prompt
2. Image-to-Video: Upload an image and provide a text prompt describing the desired animation
3. Adjust advanced settings as needed:
   • More sampling steps = higher quality but slower
   • Higher guidance scale = more adherence to prompt
   • Lower shift values (e.g., 3.0) recommended for 480p videos

Technical Details

• Frame rate: 24 FPS
• Supported resolutions: Multiples of 32 (128-1280)
• Frame count: 8-121 frames (automatically adjusted to 4n+1 format)
• Default duration: 2.0 seconds (48 frames)
• Flash Attention: Optional - automatically uses PyTorch's built-in attention if flash-attn is not available

Check out the configuration reference at https://huggingface.co/docs/hub/spaces-config-reference