HumanRF: High-Fidelity Neural Radiance Fields for Humans in Motion

TLDR

High‑fidelity human performance representation is essential for film, games, and videoconferencing, yet most work is limited to 4 MP resolution; this study tackles 12 MP synthesis. The authors introduce HumanRF to close the gap to production‑level quality by capturing full‑body appearance in motion from multi‑view video and enabling playback from novel viewpoints. HumanRF is a 4D dynamic neural scene representation that factorizes space‑time into a temporal matrix‑vector decomposition, encoding fine details at high compression from 12 MP footage collected in the ActorsHQ dataset of 160 cameras and 16 sequences. The method.

Abstract

Representing human performance at high-fidelity is an essential building block in diverse applications, such as film production, computer games or videoconferencing. To close the gap to production-level quality, we introduce HumanRF, a 4D dynamic neural scene representation that captures full-body appearance in motion from multi-view video input, and enables playback from novel, unseen viewpoints. Our novel representation acts as a dynamic video encoding that captures fine details at high compression rates by factorizing space-time into a temporal matrix-vector decomposition. This allows us to obtain temporally coherent reconstructions of human actors for long sequences, while representing high-resolution details even in the context of challenging motion. While most research focuses on synthesizing at resolutions of 4MP or lower, we address the challenge of operating at 12MP. To this end, we introduce ActorsHQ, a novel multi-view dataset that provides 12MP footage from 160 cameras for 16 sequences with high-fidelity, per-frame mesh reconstructions. We demonstrate challenges that emerge from using such high-resolution data and show that our newly introduced HumanRF effectively leverages this data, making a significant step towards production-level quality novel view synthesis.