Description:
This is an advanced Image-to-Video (I2V) workflow designed for the Wan 2.2 (14B) model. Unlike standard workflows that generate short 5-second clips, this setup utilizes SVI Pro nodes and Image Batch Extension with Overlap to create longer, seamless video sequences from a single input image. It automatically handles the "handover" between video segments to ensure consistent motion and lighting.
Key Features:
- Long-Video Generation: Uses ImageBatchExtendWithOverlap to chain multiple generations together, creating longer videos than the standard model limit.
- High Consistency: Implements specific SVI_v2_PRO LoRAs (High/Low noise) to maintain character and background stability across frames.
- Text Guidance: Allows you to prompt specific actions (e.g., "The woman is reading," "She smiles") to guide the animation.
- Upscaling & Refinement: Includes ImageScaleByAspectRatio and ImageResizeKJ to ensure input images are perfectly formatted for the model (480p/720p).
Links to my patreon and youtube: