Skip to content

Wan 2.2 SVI four-pass V3

Build an image-to-video sequence across four passes with separate LoRA controls.

All workflows
Wan 2.2Video

Wan 2.2 · image-to-video · SVI · multi-pass · LoRA · upscale · ComfyUI

Workflow preview

Wan 2.2 SVI four-pass V3 workflow overview showing node groups and connections
View raw JSON

272 KB · JSON file

Need a hand? Load your workflow or troubleshoot your Runpod setup.

  • Replace prompt placeholders and supply your own input images, video or audio before running.
  • Save wan2.2_i2v_A14b_high_noise_lora_rank64_lightx2v_4step_1022.safetensors as ComfyUI/models/loras/i2v_lightx2v_high_noise_model.safetensors.
  • Save wan2.2_i2v_A14b_low_noise_lora_rank64_lightx2v_4step_1022.safetensors as ComfyUI/models/loras/i2v_lightx2v_low_noise_model.safetensors.
  • RIFE downloads its checkpoint on first use; the listed path is the frame-interpolation node's default cache.

Models to download

Check the required model files before running this workflow.