Documentation for the LivePortraitRetargeting Node

Overview

The LivePortraitRetargeting node is a component of the ComfyUI platform tailored for retargeting features in facial manipulation tasks. Specifically, this node deals with the manipulation and retargeting of facial landmarks to produce dynamic, expressive transformations of face images. This node allows for detailed control over aspects like eye and lip retargeting, enabling advanced facial animation effects.

Functionality

The LivePortraitRetargeting node takes facial landmark data and configuration settings and processes this information to define how features such as eyes and lips should be retargeted in the resulting animations. It generates a structured set of retargeting information that can be used downstream in a workflow to influence animation or transformation effects.

Inputs

• driving_crop_info: This input provides the node with information about the cropped images of driving data (i.e., the images that will serve as the base or template for retargeting). This information includes details required to locate and manipulate facial landmarks within the driving images.

• eye_retargeting: A boolean input that specifies whether eye retargeting should be enabled. When set to true, the node will apply transformations to the eyes in the output based on the inputs provided.

• eyes_retargeting_multiplier: A floating-point input that determines the degree or intensity of eye retargeting. This value ranges from 0.01 to 10.0 and is used to scale the effects of eye retargeting related to the driving images.

• lip_retargeting: Similar to eye retargeting, this boolean input decides if the node should perform lip retargeting.

• lip_retargeting_multiplier: A floating-point number that scales the amount of lip retargeting applied. It ranges from 0.01 to 10.0, allowing fine control over lip manipulation.

Outputs

• retargeting_info: The primary output from this node is a structured dictionary containing all the processed retargeting information. This includes settings and landmark positions that define how the eyes and lips should be animated in the output images or videos. This information can be passed to other nodes in a workflow for further processing or rendering.

Use in ComfyUI Workflows

The LivePortraitRetargeting node is typically used in workflows related to facial animation and image synthesis. It is suitable for:

• Creating Animated Portraits: By adjusting the retargeting multipliers and settings, users can produce animated portraits where the eyes and lips follow movements defined by the driving images.

• Enhancing Realism: Animations requiring more lifelike expressions can benefit from the precise control over facial feature retargeting this node provides, enhancing the overall realism.

• Custom Facial Effects: Alongside other nodes in the ComfyUI system, this node can be used to design and apply custom facial effects for creative projects, digital avatars, or interactive media.

Special Features and Considerations

• Precision Control: Through detailed input settings like retargeting multipliers, the node offers precision control over the transformation effects applied to facial landmarks.

• Integration with Other Nodes: The structured output can seamlessly integrate with other nodes designed for further processing, compositing, or rendering in ComfyUI workflows.

• Configurable: Multiple boolean and floating-point inputs allow users to tailor the processing to suit different use cases and desired outcomes.

The LivePortraitRetargeting node offers powerful capabilities for the retargeting and manipulation of facial features, making it an essential tool for projects in facial animation and image transformation within the ComfyUI platform.