Real-time volumetric rendering of dynamic humans

TL;DR

This paper introduces a fast, real-time volumetric rendering method for dynamic humans from monocular videos, enabling efficient reconstruction and visualization on mobile devices with minimal quality loss.

Contribution

It presents a lightweight deformation model and a novel local ray marching rendering technique that significantly speeds up reconstruction and allows real-time visualization on mobile VR devices.

Findings

Reconstruction time reduced to less than 3 hours from 72 hours.

Achieves 40 fps rendering on mobile VR devices.

Maintains competitive visual quality with state-of-the-art methods.

Abstract

We present a method for fast 3D reconstruction and real-time rendering of dynamic humans from monocular videos with accompanying parametric body fits. Our method can reconstruct a dynamic human in less than 3h using a single GPU, compared to recent state-of-the-art alternatives that take up to 72h. These speedups are obtained by using a lightweight deformation model solely based on linear blend skinning, and an efficient factorized volumetric representation for modeling the shape and color of the person in canonical pose. Moreover, we propose a novel local ray marching rendering which, by exploiting standard GPU hardware and without any baking or conversion of the radiance field, allows visualizing the neural human on a mobile VR device at 40 frames per second with minimal loss of visual quality. Our experimental evaluation shows superior or competitive results with state-of-the art methods while obtaining large training speedup, using a simple model, and achieving real-time rendering.