Light4D: Training-Free Extreme Viewpoint 4D Video Relighting

2026-02-12

AI Summary

This work addresses the challenge of relighting 4D videos under extreme viewpoint changes, where paired training data are scarce and temporal consistency is difficult to maintain. The authors propose the first training-free framework for high-quality relighting, built upon the IC-Light architecture. By decoupling optical flow guidance to preserve geometric structure and integrating a temporally consistent attention mechanism with deterministic regularization, the method effectively suppresses flickering while ensuring lighting fidelity. Evaluated across camera rotations ranging from −90° to 90°, the approach significantly outperforms existing methods, achieving notable advances in both temporal coherence and visual quality.

Technology Category

• Computer Vision: Computational Photography, Image & Video Synthesis
• Machine Learning: Multimodal Learning
• Knowledge Representation and Reasoning: Geometric, Spatial, and Temporal Reasoning

Application Category

• Graph Algorithms and Modeling for the Web: Algorithms and analysis for heterogeneous, signed, attributed, multi-relational, temporal, higher-order, and annotated Web-related graphs
• Responsible Web: Machine-in-the-loop, human agency and autonomy
• Search and Retrieval-Augmented AI: Web learning to rank, online learning, and counterfactual learning for ranking

Abstract

Recent advances in diffusion-based generative models have established a new paradigm for image and video relighting. However, extending these capabilities to 4D relighting remains challenging, due primarily to the scarcity of paired 4D relighting training data and the difficulty of maintaining temporal consistency across extreme viewpoints. In this work, we propose Light4D, a novel training-free framework designed to synthesize consistent 4D videos under target illumination, even under extreme viewpoint changes. First, we introduce Disentangled Flow Guidance, a time-aware strategy that effectively injects lighting control into the latent space while preserving geometric integrity. Second, to reinforce temporal consistency, we develop Temporal Consistent Attention within the IC-Light architecture and further incorporate deterministic regularization to eliminate appearance flickering. Extensive experiments demonstrate that our method achieves competitive performance in temporal consistency and lighting fidelity, robustly handling camera rotations from -90 to 90. Code: https://github.com/AIGeeksGroup/Light4D. Website: https://aigeeksgroup.github.io/Light4D.

Problem

• 4D relighting
• extreme viewpoint
• temporal consistency
• training-free
• video relighting

Innovation

• training-free
• 4D relighting
• temporal consistency
• disentangled flow guidance
• extreme viewpoint