ComfyUI WAN2.1 VACE Video Helper Nodes

WORKFLOW URL: wan2_1VaceInfinite.json

This collection of custom nodes for ComfyUI is specifically designed to automate and streamline the process of creating infinite videos with WAN2.1 VACE.

Installation

Method 1: Using Git (Recommended)

1. Open a terminal or command prompt.
2. Navigate to the custom_nodes directory inside your ComfyUI installation.

   cd path/to/ComfyUI/custom_nodes
   

3. Clone the repository:

   git clone https://github.com/heheok/comfyui_wan2.1_vace_infinite_helpers.git custom_scripts
   

4. Restart ComfyUI.

---

Nodes

CyclicCharacterAndBackgroundPrompt

This node automates the generation of prompts by cycling through predefined lists, allowing for dynamic and varied video creation without manual intervention.

Inputs:

• char_prompts_json: A JSON array of strings for character prompts.
• bg_prompts_json: A JSON array of strings for background prompts.
• cam_prompts_json: A JSON array of strings for camera movement or additional prompts.

Functionality:

The node combines one prompt from each category to generate a final prompt. If the lists have different lengths, the shorter lists will loop. For example, if you have 2 character prompts and 1 background prompt, the background prompt will be repeated for each character prompt.

IMPORTANT NOTE:

This node creates and manages a state file named cyclic_prompt_indices.json in the root ComfyUI folder. This file tracks the current index for each prompt list. To reset the cycle and start from the beginning of your prompt lists, you can either:

• Delete the cyclic_prompt_indices.json file.
• Manually edit the file to reset the indices: {"char_idx": 0, "bg_idx": 0, "add_idx": 0}

LatestVideoFromFolder

This node automatically finds and loads the most recently created video file from a specified directory. This is essential for creating a continuous video loop, as it feeds the last generated video into the next generation cycle.

Inputs:

• folder_path: The absolute path to the directory containing your generated videos. For example: C:\ComfyUI\output\padded

Outputs:

• video: The loaded video file.

PrepareControlVideo

This node prepares the video input for the ControlNet, which is crucial for maintaining consistency between video segments.

Inputs:

• video_frames: The video to be prepared, typically from the VAE DECODE node.
• tail_count: The number of frames to take from the end of the input video.
• pad_count: The number of gray frames to add to the END of the generated padded video segment.

Functionality:

The node takes the last tail_count frames of the input video, puts them in the front and appends pad_count gray frames. This padded video is then used as the control video for the next segment's generation, ensuring a smooth transition.

---

Workflow & Important Considerations

Full Video vs. Padded Video

For a seamless workflow, it is recommended to save your generated videos into two separate locations:

• Full Video Path: This folder will contain the final, un-padded video segments that you will later merge together manually to create your final long-form video.
• Padded Video Path: This folder is used by the LatestVideoFromFolder node. It should contain the padded videos that are used as control inputs for subsequent generations.

Initial Generation - VERY IMPORTANT!

For the very first video generation in your sequence, you MUST DISABLE the group of nodes responsible for handling the "Previous Video to Control Video" (which includes LatestVideoFromFolder). This is because there is no preceding video to use as a control input.

After the first video has been generated, you must ENABLE this group for all subsequent generations so that the system can create a continuous, looping video.