D-NeuRA: customizable dynamic neural rendering for human avatars with disentangled pose and appearance

Ming Jiang, Jiawei Li, Yao Lu, Feng ZhangJournal of King Saud University - Computer and Information Sciences 2025 · DOI

Abstract

Neural Radiance Fields (NeRF) have become a cornerstone for generating human avatars, particularly in novel view synthesis. However, existing NeRF-based methods often rely on static features that struggle to adapt to the fine-grained variations caused by dynamic poses and changing lighting. We present D-NeuRA, a human avatar modeling framework that integrates static theme priors with image-driven dynamic features. By employing a novel Multi-Scale Bitemporal Dynamic Feature Fusion module and a Median-Enhanced Channel and Spatial Attention (MECS) mechanism, D-NeuRA achieves high-fidelity rendering with disentangled control over pose, shape, and appearance. Our approach demonstrates superior robustness and generalization in complex dynamic scenes compared to current state-of-the-art methods.

Introduction

Creating a digital avatar that moves naturally and reacts realistically to environmental changes remains a significant challenge in computer vision. While static 3D reconstruction has matured, "dynamic" realism—capturing how skin stretches, muscles flex, and fabric wrinkles during movement—is often lost in translation. Traditional NeRF models typically learn a pose-independent "base" representation. While this facilitates basic customization, it fails to capture pose-dependent details, often resulting in blurred textures or "ghosting" artifacts during complex motions.

D-NeuRA framework capabilities: from a single input image, the model enables novel pose synthesis, view changes, relighting, reshaping, retexturing, and reshadowing.

Methodology

D-NeuRA addresses these limitations through three primary technical innovations designed to bridge the gap between static stability and dynamic realism.

1. Adaptive Appearance Feature Construction

Rather than relying solely on a pre-learned static latent code, D-NeuRA extracts dynamic features directly from input images. The system maps 3D sampling points to 2D image coordinates, using bilinear sampling to capture local texture and pose information. These dynamic features are then fused with static "theme" features via a Multi-Scale Bitemporal Dynamic Feature Fusion module, ensuring the avatar maintains global consistency while reflecting local, pose-specific details.

2. Two-Stage Pose Enhancement

To refine geometric modeling, the framework employs a two-stage strategy:

  • Stage 1: Key Pose Retrieval: The model retrieves and interpolates features from the five most similar poses in the training set to provide a geometric baseline.
  • Stage 2: MECS Module: To prevent the "averaging blur" common in interpolation, the Median-Enhanced Channel and Spatial Attention (MECS) module is introduced. By incorporating median pooling—which is inherently robust to outliers—the model preserves sharp geometric cues, such as deep clothing folds, that standard mean or max pooling often smooth out.

3. Disentangled Rendering Pipeline

The framework utilizes separate decoders for SDF (geometry), Albedo (base color), and Shadow (lighting and occlusion). This disentanglement is crucial for customization, allowing users to independently modify an avatar’s body shape, clothing texture, or lighting environment without degrading the underlying model integrity.

Experimental Results

The researchers evaluated D-NeuRA on the ZJU-MoCap (multi-view) and MonoCap (monocular) benchmarks.

Quantitative Performance: D-NeuRA consistently outperformed baselines such as Neural Body (NB) and NECA. On the ZJU-MoCap dataset, it achieved a PSNR of 29.3 and a significantly lower LPIPS (0.062), indicating superior perceptual similarity to ground-truth images.

Qualitative Improvements: The model showed marked improvements in rendering complex areas like facial contours and hand gestures. In "novel pose" synthesis—rendering positions not seen during training—D-NeuRA maintained structural integrity where competing models often suffered from limb collapse or texture blurring. Furthermore, the disentangled pipeline allowed for seamless "reshadowing," where the model realistically updates shadows based on pose changes even under static external lighting.

Conclusion and Future Work

D-NeuRA represents a significant step toward fully customizable, high-fidelity digital humans. By combining the stability of static priors with the detail of image-driven dynamic features, it achieves a level of realism previously difficult to reach in real-time rendering.

Currently, the model's accuracy is linked to the quality of the underlying SMPL (Skinned Multi-Person Linear) pose estimation. Future research will focus on developing "shared representations" to allow the model to generalize across different subjects without requiring per-person optimization, moving the industry closer to a universal human rendering engine.

Summary updated

Abstract

Neural Radiance Fields (NeRF) has become a fundamental method for generating human avatars due to its successful application in novel view synthesis. Existing NeRF-based methods build static features that are independent of pose to learn and represent human albedo and shadow information, combined with external lighting information to generate fully customizable human avatars. However, these static features can only provide global, pose-independent information and are unable to adapt to the detailed variations caused by pose and lighting changes. To address this, we propose a human avatar modeling framework, D-NeuRA, which integrates static theme priors and image-driven dynamic features. Specifically, we map three-dimensional sampling points to image coordinates and perform bilinear sampling on image feature maps to extract dynamic human appearance features. We also design a Multi-Scale Bitemporal Dynamic Feature Fusion module to effectively combine dynamic features with static theme features to generate adaptive appearance features. To further enhance the expression of pose modeling at the local geometric level, we introduce a two-stage pose feature extraction and enhancement strategy. In the first stage, key pose retrieval and interpolation are used to sample and extract preliminary pose features along three axes. In the second stage, a Median-Enhanced Channel and Spatial Attention (MECS) module refines and enhances the pose features. Finally, we use multiple decoders to combine the adaptive appearance features and enhanced pose features to generate neural field properties, rendering high-quality, fully customizable dynamic human avatars. Extensive experiments show that the proposed method outperforms existing mainstream approaches in rendering realism, pose consistency, and detail preservation, particularly demonstrating stronger robustness and generalization ability in complex dynamic scenes.
BibTeX
@article{jiang2025d,
  author = {Ming Jiang and Jiawei Li and Yao Lu and Feng Zhang},
  title = {{D-NeuRA: customizable dynamic neural rendering for human avatars with disentangled pose and appearance}},
  journal = {Journal of King Saud University - Computer and Information Sciences},
  year = {2025},
  doi = {10.1007/s44443-025-00262-5},
}
Figure