README.md

metadata

title: WAN 2.2 3-Step V2V Pipeline
emoji: 🎬
colorFrom: purple
colorTo: blue
sdk: gradio
sdk_version: 5.44.1
python_version: '3.10'
app_file: app.py
pinned: false
short_description: I2V + T2V + 3-Step V2V (SAM2 → Composite → VACE)
models:
  - facebook/sam2.1-hiera-large
  - google/umt5-xxl
  - Kijai/WanVideo_comfy
  - linoyts/Wan2.2-T2V-A14B-Diffusers-BF16
  - lkzd7/WAN2.2_LoraSet_NSFW
  - r3gm/RIFE
  - TestOrganizationPleaseIgnore/WAMU_v2_WAN2.2_I2V_LIGHTNING
  - Wan-AI/Wan2.1-VACE-14B-diffusers
  - Wan-AI/Wan2.2-T2V-A14B-Diffusers
  - zerogpu-aoti/Wan2

WAN 2.2 Multi-Task Video Generation

Features

• I2V: Image-to-Video (Lightning 14B, 6-step, FP8+AoT)
• T2V: Text-to-Video (Lightning 14B, 4-step, Lightning LoRA)
• V2V: 3-Step Video-to-Video Pipeline
  1. SAM2 Segmentation: Click points on first frame → auto-track through video → mask video
  2. Composite + GrowMask: Original + mask → expanded mask + composite video (automatic)
  3. VACE Generation: Composite + grown mask + reference image + prompt → final video

V2V Workflow

Based on ComfyUI workflows: sam_optimized, sam2.1_optimized, vace_optimized