Rendering · Lumen-1
Lumen.Faces thatfeel present.
Lumen renders the persona: photoreal faces with micro-expressions, lip sync and, in the Aarohi-1 preview, full-body posture and gesture, at video frame rates.
- Output
- 1080p, 30 fps, streamed over WebRTC
- Lip sync
- Phoneme-accurate across 40+ languages
- Body
- Full-body gesture & posture (preview)
- Training
- 2 minutes of consented video for a custom face
Inside the model
What it doesfor a conversation.
- Micro-expression
- Listening is visible: small nods, a raised brow, eyes that track. Lumen animates the in-between moments that make a face feel attentive.
- Full body, in preview
- Aarohi-1 adds posture and hand gesture driven by the conversation, for use cases like yoga, fitness and presenting where the body is the message.
- Consent first
- Custom faces require verified, recorded consent from the person depicted, and every persona discloses it is an AI.
Other models
ManavOS · Lumen