ComfyUI LTXV Condition Replace

A ComfyUI custom node for combining LTXV conditioning latents from different sources.

Description

This extension provides the LTXVConditionReplace node, which combines two latent inputs:

• control_unit_latent: From LTXAddVideoICLoRAGuide (handles control unit input)
• key_frames_latent: From LTXVAddGuideMulti (handles multiple key frame images)

The node replaces control unit latent frames with key frame latent frames at corresponding positions, outputting a unified condition latent.

Installation

1. Clone this repository to your ComfyUI custom_nodes directory:

   cd ComfyUI/custom_nodes
   git clone <this-repo-url> Comfyui-LTXCondReplace
   

2. Restart ComfyUI

Usage

LTXVConditionReplace Node

Inputs:

• control_unit_latent: Latent from LTXAddVideoICLoRAGuide
• key_frames_latent: Latent from LTXVAddGuideMulti
• strength: Strength of the replacement (0.0-1.0)
  • 1.0 = full replacement
  • 0.0 = no replacement

Outputs:

• unified_control_latent: Unified condition latent combining both inputs

Example Workflow

1. Load your base video latent
2. Add key frames using LTXVAddGuideMulti → produces key_frames_latent
3. Add control unit using LTXAddVideoICLoRAGuide → produces control_unit_latent
4. Combine both using LTXVConditionReplace → produces unified_control_latent
5. Use the unified latent for sampling

Nodes

LTXVConditionReplace

Combines two latent inputs into a unified condition latent by replacing control unit latent frames with key frame latent frames.

License

See LICENSE file.