FaceSnap: Real-Time Personalized Lightstage Facial Performance Capture
2026-08-31 • Computer Vision and Pattern Recognition
Computer Vision and Pattern Recognition
AI summaryⓘ
The authors present FaceSnap, a new method that simplifies creating detailed digital faces by using just one camera instead of many. They first build a personalized 3D face model using multiple views and movement, which helps capture realistic facial details quickly afterward with only a single camera shot. FaceSnap can recreate high-quality facial geometry and textures in real-time, matching the accuracy of more complex multi-camera setups. They also introduce Multi4D, a new public benchmark to fairly compare different 3D facial capture techniques.
LightstageFacial captureMulti-view optimizationMonocular camera3D geometryTexture upscalingReal-time captureFacial performance4D reconstructionBenchmark dataset
Authors
Rukhshanda Hussain, Noé Artru, Emeline Got, Luiz Gustavo Hafemann, Alexandre Messier, Brandon Dearlove, Rafael M. O. Cruz, Abdallah Dib, Eric Granger
Abstract
Lightstage facial capture produces production-quality digital humans, but it is resource and labor-intensive. Multi-camera setups, hours of computation, and massive data storage create bottlenecks that hinder iterative workflows. This paper introduces FaceSnap, an end-to-end framework that streamlines capture via a two-stage approach. First, a one-time multi-view optimization from a range-of-motion sequence builds a personalized model encoding both geometry and expression-dependent appearance. This model then enables high-fidelity real-time facial performance capture from a single monocular lightstage camera, with no further multi-view capture required. FaceSnap jointly estimates geometry and dynamic 4K texture at 83 fps. The 4K texture is produced by a novel personalized residual upscaler that recovers subject-specific high-frequency detail, which generic upscalers fail to capture. FaceSnap achieves geometric accuracy competitive with full per-frame multi-view optimization while outperforming feed-forward methods trained on production-quality 3D data, all from a single camera view. Finally, we introduce Multi4D, a public benchmark for evaluating 4D facial reconstruction methods in lightstage environments, enabling topology-invariant geometric comparison across methods.