AI Paper Atlas

Research map / 3D vision / Gaussian splatting

Deblur4DGS: 4D Gaussian Splatting from Blurry Monocular Video

Renlong Wu, Zhilu Zhang, Mingyang Chen, Zifei Yan, Wangmeng Zuo

AAAI 2026 · 3D vision › Gaussian splatting · approach: Deformable 4D Gaussian splatting · 19 citations (OpenAlex)

Read the paperCheck your idea against 137,217 papers

AI-extracted. The measured effects below were read by AI from the paper's full text (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.

Measured effects reported in the paper

What was changedWhereMeasuredEffect
add continuous camera pose and dynamic Gaussian estimation (ECP+EDG) (ablation)Deblur4DGS · Stereo Blur dynamic areas (288x512)PSNR: 22.10 -> 22.36Improved
add continuous camera pose and dynamic Gaussian estimation (ablation)Deblur4DGS · Stereo Blur dynamic areas (720x1080)PSNR: 22.34 -> 22.63Improved
add EDG on top of ECP (ablation)Deblur4DGS · Stereo Blur dynamic areas (720x1080)PSNR: 22.39 -> 22.63Improved
add regularization terms Le, Lmfc, Lmrc (ablation)Deblur4DGS · Stereo Blur dynamic areas (288x512)PSNR: 21.87 -> 22.36Improved
add Lmrc on top of Le and Lmfc (ablation)Deblur4DGS · Stereo Blur dynamic areas (720x1080)PSNR: 22.49 -> 22.63Improved
use variable canonical Gaussians instead of a single one (ablation)Deblur4DGS · Stereo Blur dynamic areas (288x512)PSNR: 22.13 -> 22.36Improved
use blur-aware selection instead of uniform canonical Gaussians (ablation)Deblur4DGS · Stereo Blur dynamic areas (720x1080)PSNR: 22.57 -> 22.63Improved
use optical-flow-based canonical selection instead of uniform (ablation)Deblur4DGS · Stereo Blur dynamic areasPSNRNo clear effect
use Deblur4DGS vs SplineGS on BARD-GS dataDeblur4DGS · BARD-GS real-world data NVSPSNR: 23.93 -> 23.10Hurt
use Deblur4DGS vs BARD-GSDeblur4DGS · Stereo Blur synthetic NVS (288x512)PSNR: 26.91 -> 27.66Improved

Later papers that compared with this method

2 later accepted papers report Deblur4DGS as a baseline in their results tables; 2 of them report a better number on most of the results they share with it.

Earlier methods this paper compares with

Named baselines from earlier accepted papers in this paper's results tables: better than 2 of 3 on most shared results.

Matched by method name in results tables read by AI. Authors' own tables tend to favour the newer method, and a comparison can be missed when a method is renamed.

Closest papers by meaning

Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.

Open scime draft

Get the next scime Dispatch and a monthly field update

Free. We send a confirmation link first; every email has a one-click unsubscribe. Privacy