Learning Neural Exposure Fields for View Synthesis

2025-10-09

AI Summary

To address the degradation of novel-view synthesis quality in neural scene representations under exposure variations—particularly in high-dynamic-range (HDR) real-world scenes such as indoor-outdoor hybrids or windowed interiors—this paper proposes Neural Exposure Fields (NExF), the first end-to-end framework jointly modeling scene geometry, appearance, and per-point exposure parameters in 3D space. NExF embeds a learnable exposure field into a neural radiance field and employs a neural conditional mechanism to co-optimize exposure and radiance, requiring neither multi-exposure inputs nor post-processing. Evaluated on multiple real-world HDR benchmarks, NExF achieves over 55% PSNR improvement over state-of-the-art methods, with higher training efficiency and significantly enhanced 3D-consistent novel-view synthesis under complex, spatially varying illumination.

Technology Category

• Computer Vision: Computational Photography, Image & Video Synthesis
• Intelligent Robots: Multimodal Perception & Sensor Fusion
• Natural Language Processing: Language Grounding & Multi-modal NLP

Application Category

• Search and Retrieval-Augmented AI: Web learning to rank, online learning, and counterfactual learning for ranking
• User Modeling, Personalization and Recommendation: Fairness-aware retrieval and ranking
• Graph Algorithms and Modeling for the Web: Graph neural networks and deep learning approaches for Web-related graphs

Abstract

Recent advances in neural scene representations have led to unprecedented quality in 3D reconstruction and view synthesis. Despite achieving high-quality results for common benchmarks with curated data, outputs often degrade for data that contain per image variations such as strong exposure changes, present, e.g., in most scenes with indoor and outdoor areas or rooms with windows. In this paper, we introduce Neural Exposure Fields (NExF), a novel technique for robustly reconstructing 3D scenes with high quality and 3D-consistent appearance from challenging real-world captures. In the core, we propose to learn a neural field predicting an optimal exposure value per 3D point, enabling us to optimize exposure along with the neural scene representation. While capture devices such as cameras select optimal exposure per image/pixel, we generalize this concept and perform optimization in 3D instead. This enables accurate view synthesis in high dynamic range scenarios, bypassing the need of post-processing steps or multi-exposure captures. Our contributions include a novel neural representation for exposure prediction, a system for joint optimization of the scene representation and the exposure field via a novel neural conditioning mechanism, and demonstrated superior performance on challenging real-world data. We find that our approach trains faster than prior works and produces state-of-the-art results on several benchmarks improving by over 55% over best-performing baselines.

Problem

• Reconstructing 3D scenes robustly from varying exposure images
• Optimizing exposure per 3D point for consistent appearance
• Enabling accurate view synthesis in high dynamic range scenarios

Innovation

• Learns neural field predicting exposure per 3D point
• Optimizes exposure jointly with neural scene representation
• Enables high dynamic range view synthesis without post-processing