CANDLE: Illumination-Invariant Semantic Priors for Color Ambient Lighting Normalization

TL;DR

CANDLE introduces a novel illumination-invariant semantic prior using DINOv3 features and multi-layer guidance to improve color normalization under complex lighting conditions, demonstrating significant performance gains.

Contribution

The paper proposes CANDLE, a new method leveraging DINOv3 features and multi-layer guidance for robust color ambient lighting normalization, outperforming prior approaches.

Findings

Achieved +1.22 dB PSNR gain over previous methods.

Secured 3rd place in NTIRE 2026 ALN Challenge.

Obtained 2nd place in fidelity with lowest FID on White Lighting track.

Abstract

Color ambient lighting normalization under multi-colored illumination is challenging due to severe chromatic shifts, highlight saturation, and material-dependent reflectance. Existing geometric and low-level priors are insufficient for recovering object-intrinsic color when illumination-induced chromatic bias dominates. We observe that DINOv3's self-supervised features remain highly consistent between colored-light inputs and ambient-lit ground truth, motivating their use as illumination-robust semantic priors. We propose CANDLE (Color Ambient Normalization with DINO Layer Enhancement), which introduces DINO Omni-layer Guidance (D.O.G.) to adaptively inject multi-layer DINOv3 features into successive encoder stages, and a color-frequency refinement design (BFACG + SFFB) to suppress decoder-side chromatic collapse and detail contamination. Experiments on CL3AN show a +1.22 dB PSNR gain over the strongest prior method. CANDLE achieves 3rd place on the NTIRE 2026 ALN Color Lighting Challenge and 2nd place in fidelity on the White Lighting track with the lowest FID, confirming strong generalization across both chromatic and luminance-dominant illumination conditions. Code is available at https://github.com/ron941/CANDLE.