Face
Create a facial identity asset from a reference photo for lip-sync and identity preservation.
Overview
The Face node creates a facial identity asset from a clear reference photo. It preserves the subject’s facial identity in generated images and is primarily used for lip-sync operations and head replacement in video compositions. A clear, front-facing photo is required as input for best results.
Configuration
| Field | Type | Default | Description |
|---|---|---|---|
| Name | string | "" |
Name for this face asset. |
| Description | string | "" |
Optional text description of facial features or context. |
| Style | enum | "realistic" |
Visual style. Options: realistic, anime, 3d-pixar, illustration. |
| Reference Photo | image URL | required | A clear face photo. This is the primary input – the photo should show a front-facing view with good lighting and minimal occlusion. |
Inputs & Outputs
Inputs:
in– Source image containing the face (typically from an Upload Image node).
Outputs:
faceRef– Face reference for use in lip-sync nodes, scene nodes, and other identity-preserving operations.Best Practices
- Use a clear, well-lit, front-facing photo with the face clearly visible and unobstructed.
- Avoid photos with heavy makeup, sunglasses, or extreme angles, as these reduce identity accuracy.
- One face per node – create separate Face nodes for different people.
- Generate the face asset before connecting it to downstream lip-sync or scene nodes.
Common Use Cases
- Providing facial identity for lip-sync video generation.
- Maintaining face consistency across multiple generated images in a workflow.
- Creating a reusable face identity for talking-head video content.
- Preserving a speaker’s identity when generating video from audio narration.
Tips
- The face reference carries enough identity information for downstream nodes to maintain visual consistency without re-uploading the photo.
- Face assets are persisted in the project database and can be reused across workflows within the same project.
- Unlike the Character node, the Face node focuses solely on facial identity rather than full-body variations.
- For best lip-sync results, use a photo where the subject’s mouth is clearly visible and in a neutral position.