Real-Time Radiance Fields for Single-Image Portrait View Synthesis

TL;DR

This paper introduces a fast, real-time method to generate high-quality 3D portrait views from a single image using a neural radiance field, trained solely on synthetic data.

Contribution

It presents a novel Vision Transformer-based encoder and training strategy that distills 3D GAN knowledge into a single-shot, real-time portrait synthesis system.

Findings

Achieves 24 fps rendering on consumer hardware.

Outperforms GAN-inversion baselines in quality.

Demonstrates robustness and high quality in real-world portraits.

Abstract

We present a one-shot method to infer and render a photorealistic 3D representation from a single unposed image (e.g., face portrait) in real-time. Given a single RGB input, our image encoder directly predicts a canonical triplane representation of a neural radiance field for 3D-aware novel view synthesis via volume rendering. Our method is fast (24 fps) on consumer hardware, and produces higher quality results than strong GAN-inversion baselines that require test-time optimization. To train our triplane encoder pipeline, we use only synthetic data, showing how to distill the knowledge from a pretrained 3D GAN into a feedforward encoder. Technical contributions include a Vision Transformer-based triplane encoder, a camera data augmentation strategy, and a well-designed loss function for synthetic data training. We benchmark against the state-of-the-art methods, demonstrating significant improvements in robustness and image quality in challenging real-world settings. We showcase our results on portraits of faces (FFHQ) and cats (AFHQ), but our algorithm can also be applied in the future to other categories with a 3D-aware image generator.