ComfyUI Extension: WanVideoLooper

Authored by SquirrelRat

Created 8 days ago

Updated about 24 hours ago

0 stars

This is a set of custom nodes for ComfyUI designed for generating sequential long video clips based on the Wan 2.2 model architecture, handling continuity between segments and offering advanced control over the sampling process

Custom Nodes (0)

README

WanVideoLooper Nodes for ComfyUI

This is a set of custom nodes for ComfyUI designed for generating sequential long video clips, handling continuity between segments, and offering advanced control over the sampling process, including per-segment LoRA/model application.

This pack contains three nodes:

WanVideo Looper: The main node that takes base models, a list of prompts, and various settings to generate a sequence of video clips, merging them seamlessly.
WanVideo Looper Prompts: A new helper node for managing multiline positive prompts, a global negative prompt, prefixes/suffixes, and saving/loading prompt profiles.
WanVideo Looper Lora Sequencer: A helper node designed to feed pre-patched Model and CLIP objects into the WanVideo Looper for specific segments, enabling per-segment LoRA application.

Watch the Tutorial Video (Original): (Note: This video may be for an older version but can still be helpful for understanding the core concepts.)

How to Install

Navigate to your ComfyUI installation directory.
Go into the ComfyUI/custom_nodes/ folder.
Create a new folder inside custom_nodes named WanVideoLooper.
Place all the .py files inside the new WanVideoLooper folder:
- __init__.py
- wan_looper_node.py
- wan_lora_sequencer.py
- wan_prompts_node.py
Create a js subfolder inside your new folder: ComfyUI/custom_nodes/WanVideoLooper/js.
Place all the .js files inside that js folder:
- wan_lora_sequencer.js
- wan_prompts_node.js
(Required for Color Match) Open a terminal or command prompt, activate your ComfyUI virtual environment (if you use one), and run:
```
pip install color-matcher
```
Restart your ComfyUI server.

The nodes will be available by right-clicking the canvas and selecting Add Node > WanVideoLooper.

1. WanVideo Looper

This is the core node for generating video sequences segment by segment. It intelligently handles continuity and uses an advanced MoE-style sampling approach.

Features

Prompt List Input: Takes a prompt_list and negative_prompt directly from the WanVideo Looper Prompts node.
Seamless Continuity: Uses frame merging (frame_merge) to ensure smooth transitions between segments.
MoE-Style Sampling: Integrates Mixture of Experts logic with a single steps input and automatic calculation of the split point (model_switch_point).
Internal Color Matching: Can apply MKL color matching between segments (color_match) or match the entire video to the very last frame (color_match_lastframe) to fix color drift. Requires pip install color-matcher.
Per-Segment Models/LoRAs: Supports an optional input from the WanVideo Lora Sequencer to use different, pre-patched models for specific segments.
Dry Run Mode: Includes an enable_dry_run toggle to quickly check setup and logs without performing sampling/decoding.

Inputs & Outputs

Required Inputs
- model_high: The BASE high-noise model.
- model_low: The BASE low-noise model.
- clip: The BASE CLIP model.
- vae: The VAE model used for decoding.
- start_image: The seed image for the first loop.
- positive_prompts: Connect the prompt_list output from the WanVideo Looper Prompts node.
- negative_prompt: Connect the negative_prompt output from the WanVideo Looper Prompts node.
- seed: The random seed.
- steps: The total number of sampling steps for each loop.
- enable_motion_cfg: Enable a separate CFG for the very first step.
- cfg_motion_step: CFG for the first step ONLY.
- cfg_high_noise: CFG for the high-noise model (structure/motion).
- cfg_low_noise: CFG for the low-noise model (details).
- sampler_name: The sampling algorithm.
- scheduler: The noise scheduler.
- model_switch_point: Timestep to switch from high-noise to low-noise model (e.g., 0.9).
- sigma_shift: Applies sigma shift to both models (e.g., 8.0).
- denoise: Denoise amount (1.0 = full new image).
- width/height: The output video dimensions.
- frame_merge: Number of frames to overlap/merge between loops.
- duration_sec: Duration of each segment in seconds (5s recommended).
- color_match: (New!) Enable segment-to-segment MKL color matching.
- color_match_lastframe: (New!) Override color_match and instead match all frames to the very last frame of the sequence.
- color_match_strength: (New!) Strength of the color match (0.0 to 1.0).
- enable_dry_run: Skips sampling and decoding to quickly check setup.
Optional Inputs
- clip_vision: (Optional) A CLIP Vision model for guiding the start image.
- model_clip_sequence: (Optional) Connect the WanVideo Lora Sequencer output here.
- color_match_ref: (New!) (Optional) A static reference image for color matching. If omitted, uses the last frame of the previous segment.
Outputs
- images: The final, concatenated batch of all generated video frames.
- last_frame: The single, very last frame of the entire sequence.

2. WanVideo Looper Prompts

A helper node to manage all your prompts, prefixes, and profiles in one place.

Features

Multiline Prompts: Enter one positive prompt per line. Each line becomes a new video segment. Empty lines are ignored.
Global Negative Prompt: A single multiline text box for your negative prompt.
Prefix/Suffix: Automatically add text (like masterpiece,) before and after every single prompt (positive and negative) with a toggle to enable or disable.
Profile Management: Save your complex prompt setups (positive, negative, prefix, suffix) to named profiles in your browser's localStorage.
Save/Load/Delete: Easily save new profiles, load existing ones from a dropdown, and delete old profiles.

Inputs & Outputs

Inputs (Widgets)
- prompts: Multiline text for positive prompts (one per line).
- negative_prompt: Multiline text for the global negative prompt.
- enable_prefix_suffix: A "true"/"false" toggle.
- prefix: Text to add before each prompt.
- suffix: Text to add after each prompt.
- (Profile controls for saving/loading)
Outputs
- prompt_list: A list of processed positive prompts, ready for the main looper node.
- negative_prompt: The single, processed negative prompt string.

3. WanVideo Looper Lora Sequencer

This helper node allows you to specify different, pre-patched Model and CLIP objects for each segment of your video sequence. It's the key to applying different LoRAs to different parts of your video.

Features

Segment Inputs: Provides 10 sets of optional inputs (model_high_X, model_low_X, clip_X) corresponding to the first 10 prompt segments.
Dynamic Inputs: The node's UI is dynamic. It only shows inputs for segments that are connected. Connect segment 1 to see segment 2, and so on.
Clean Workflow: You patch your models using standard ComfyUI nodes (Load LoRA, etc.) outside the sequencer. Connect the final patched MODEL and CLIP into the corresponding inputs on this node.
Bundles Inputs: The sequencer bundles these connected models/clips into a list for the main looper.
Fallback: If you leave the inputs for a specific segment unconnected, the WanVideo Looper will automatically fall back to using the clean base models for that segment.

Inputs & Outputs

Optional Inputs (30 total)
- model_high_1 to model_high_10: The pre-patched HIGH model for each segment.
- model_low_1 to model_low_10: The pre-patched LOW model for each segment.
- clip_1 to clip_10: The pre-patched CLIP