Overview

Video restoration studies how to recover useful visual information from degraded footage, including motion blur, low resolution, noise, compression artifacts, weather, flicker, and damaged archival film. Methods use temporal alignment and fusion, recurrent networks and transformers, physical models of how degradation forms, and diffusion or flow-matching generative priors; some also draw on metadata, reference frames, or other sensors. Evaluation spans settings such as super-resolution, deblurring, HDR reconstruction, and object removal, using benchmarks such as REDS, GoPro, and VFHQ alongside real-world footage. A central challenge is balancing perceptual detail and realistic generation with fidelity to the original scene, temporal stability, and the speed and memory needed for long or high-resolution video.

Latest papers

CardsList

63 papers5 papers in the last four weeks, down 50% on the four weeks before. 0.0% of all new papers.

  1. F2HDR: Two-Stage HDR Video Reconstruction via Flow Adapter and Physical Motion Modeling

    Mar 16, 2026Huanjing Yue, Dawei Li, Shaoxiong Tu +1HDR Video ReconstructionHDR Imaging

  2. RobustSCI: Beyond Reconstruction to Restoration for Snapshot Compressive Imaging under Real-World Degradations

    Mar 8, 2026Hao Wang, Zhankuo Xu, Jiong Ni +3Image RestorationImage Corruption Robustness

  3. FMA-Net++: Motion- and Exposure-Aware Joint Video Super-Resolution and Deblurring

    Dec 4, 2025Geunhyuk Youk, Jihyong Oh, Munchurl KimImage RestorationVideo Super-Resolution