EUVIMEDEUVIMED entdecken
Open Access · CC BY 4.0

RAWVAR: Visual Autoregressive Modeling for Color-Faithful Low-Light RAW Image Restoration

Jiyuan Xia, Panjun Liu, Yuanshen Guan, Yong Li, Yinglong Li, Zhiqiang Lang, Ruikang Xu, Chang Chen, Xiongchao Chen, Zhengjun Zha, Fenglong Song, Zhiwei Xiong

ACM Transactions on Graphics · 09.11.2026 · 45 (6) · 1-15

DOI: 10.1145/3842554 ↗

Kurz gefasst

Fachartikel in der Zeitschrift ACM Transactions on Graphics, online erschienen am 09.11.2026. Verfasst von Jiyuan Xia und 11 weiteren Autor:innen. Die Arbeit nennt 65 Literaturangaben. Der Volltext ist frei zugänglich (Open Access, CC BY 4.0).

AbstractBibliografische AngabenZitierenVerwandte Studien

Abstract (Originalsprache)

Visual autoregressive (VAR) modeling provides a promising generative solution for RGB image restoration, owing to its strong ability to generate realistic details and recover rich scene information. Nevertheless, directly applying VAR to low-light RAW image restoration still presents two main challenges. First, RGB-pretrained VAE tokenizers are mismatched with RAW sensor statistics, causing color bias after domain transfer. Second, VAR requires tiled inference for high-resolution RAW images, where independent token prediction across tiles may introduce color discrepancies that are further amplified by the Image Signal Processing (ISP) pipeline. To address these issues, we propose RAWVAR, the first visual autoregressive framework for low-light RAW image restoration. For effective domain transfer, we design a Cross-Domain Codebook (CDC), which freezes the RGB-pretrained encoder and original codebook while appending a trainable RAW-domain codebook, forming a cross-domain latent space for texture-preserving and color-faithful RAW representation. To mitigate color inconsistency during tiled inference, we further design a Position-aware Token Alignment (PTA) strategy. During coarse-to-fine prediction, PTA aligns the token logits of semantically corresponding regions across tiles, ensuring consistent predictions for shared content. Furthermore, a Global Color Refinement (GCR) module globally refines the discrete latents to enhance image-wise color consistency. Compared with diffusion-based methods, RAWVAR provides a more efficient generative restoration pipeline by avoiding iterative denoising. Extensive experiments show that RAWVAR produces color-faithful and perceptually superior results, outperforming existing generative restoration methods.

Zitieren

APA

Xia, J., Liu, P., Guan, Y., Li, Y., Li, Y., Lang, Z., Xu, R., Chen, C., Chen, X., Zha, Z., Song, F., & Xiong, Z. (2026). RAWVAR: Visual Autoregressive Modeling for Color-Faithful Low-Light RAW Image Restoration. ACM Transactions on Graphics, 45(6), 1-15. https://doi.org/10.1145/3842554

Vancouver

Xia J, Liu P, Guan Y, Li Y, Li Y, Lang Z, et al. RAWVAR: Visual Autoregressive Modeling for Color-Faithful Low-Light RAW Image Restoration. ACM Transactions on Graphics. 2026;45(6):1-15. doi:10.1145/3842554

BibTeX
@article{Xia2026,
  author = {Xia, Jiyuan and Liu, Panjun and Guan, Yuanshen and Li, Yong and Li, Yinglong and Lang, Zhiqiang and Xu, Ruikang and Chen, Chang and Chen, Xiongchao and Zha, Zhengjun and Song, Fenglong and Xiong, Zhiwei},
  title = {RAWVAR: Visual Autoregressive Modeling for Color-Faithful Low-Light RAW Image Restoration},
  journal = {ACM Transactions on Graphics},
  year = {2026},
  volume = {45},
  number = {6},
  pages = {1--15},
  doi = {10.1145/3842554}
}
Kostenlos starten →