Get the latest tech news

URAvatar: Universal Relightable Gaussian Codec Avatars

We first employ a large relightable corpus of multi-view facial performances to train a cross-identity decoder that can generate volumetric avatar representations. Then given a single phone scan of an unseen identity, we reconstruct the head pose, geometry, and albedo texture, and fine-tune our pretrained relightable prior model. Our final model provides disentangled control over relighting, gaze and neck control.

Get the Android app

Or read this on Hacker News