Portrait4D: Learning One-Shot 4D Head Avatar Synthesis using Synthetic Data
This work addresses the problem of realistic 4D head synthesis for applications like virtual reality, but it is incremental as it builds on existing generative and transformer-based approaches.
The paper tackles the challenge of one-shot 4D head avatar synthesis by learning from large-scale synthetic data, achieving superior performance over prior methods as demonstrated in experiments.
Existing one-shot 4D head synthesis methods usually learn from monocular videos with the aid of 3DMM reconstruction, yet the latter is evenly challenging which restricts them from reasonable 4D head synthesis. We present a method to learn one-shot 4D head synthesis via large-scale synthetic data. The key is to first learn a part-wise 4D generative model from monocular images via adversarial learning, to synthesize multi-view images of diverse identities and full motions as training data; then leverage a transformer-based animatable triplane reconstructor to learn 4D head reconstruction using the synthetic data. A novel learning strategy is enforced to enhance the generalizability to real images by disentangling the learning process of 3D reconstruction and reenactment. Experiments demonstrate our superiority over the prior art.