1. Workflow Overview

This is a Wan2.1-based video generation workflow specialized in creating object-splitting visual effects (e.g., explosions, fragmentation). Key features:

• Generates dynamic videos from a single image + text prompt

• Supports LoRA fine-tuning (e.g., "Division" or "Bloom" effects)

• Optional post-processing: upscaling + frame interpolation (FILM VFI)

2. Core Models

3. Key Nodes

4. Workflow Structure

Group 1: Video Generation

• Inputs: Image + prompt (e.g., "object shattering into pieces")

• Output: 480P raw video

• Key Params: Frames (default 81), resolution (768xauto)

Group 2: Post-Processing

• Functions: 4x upscale + 10x frame interpolation

• Warning: Reduce original frames/resolution to avoid OOM

5. Inputs & Outputs

Input Parameters:

• positive_prompt: Describes splitting effect (e.g., "woman spinning and exploding")

• num_frames: Total frames (recommend ≤81)

• generation_width: Video width (recommend ≤768px)

Output:

• MP4 video (H.264, 16FPS by default)

6. Notes

1. VRAM: Minimum 12GB GPU (for 480P generation)

2. Common Errors:

   • module 'comfy.model_management' → Insufficient VRAM, reduce frames

   • Interpolation failure → Disable FILM VFI or lower multiplier

3. LoRA Switching: Use WanVideoLoraSelect node to change effects