Deformable Model-Driven Neural Rendering for High-Fidelity 3D Reconstruction of Human Heads Under Low-View Settings

TL;DR

This paper introduces a geometry decomposition and two-stage training approach for high-fidelity 3D head reconstruction from limited views, effectively capturing details without 3D supervision.

Contribution

It proposes a novel zero level-set representation with a coarse-to-fine training strategy that improves 3D head reconstruction under low-view conditions.

Findings

Outperforms existing methods in accuracy and view synthesis

Effective geometry decomposition enhances detail capture

Pre-trained templates generalize to unseen individuals

Abstract

Reconstructing 3D human heads in low-view settings presents technical challenges, mainly due to the pronounced risk of overfitting with limited views and high-frequency signals. To address this, we propose geometry decomposition and adopt a two-stage, coarse-to-fine training strategy, allowing for progressively capturing high-frequency geometric details. We represent 3D human heads using the zero level-set of a combined signed distance field, comprising a smooth template, a non-rigid deformation, and a high-frequency displacement field. The template captures features that are independent of both identity and expression and is co-trained with the deformation network across multiple individuals with sparse and randomly selected views. The displacement field, capturing individual-specific details, undergoes separate training for each person. Our network training does not require 3D supervision or object masks. Experimental results demonstrate the effectiveness and robustness of our geometry decomposition and two-stage training strategy. Our method outperforms existing neural rendering approaches in terms of reconstruction accuracy and novel view synthesis under low-view settings. Moreover, the pre-trained template serves a good initialization for our model when encountering unseen individuals.