AI Paper Atlas

Research map / 3D vision / Gaussian splatting

SplatSuRe: Selective Super-Resolution for Multi-view Consistent 3D Gaussian Splatting

Pranav Asthana, Alex Hanson, Allen Tu, Tom Goldstein, Matthias Zwicker, Amitabh Varshney

CVPR 2026 · 3D vision › Gaussian splatting · approach: Regularization and primitive design for 3DGS · 5 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
apply SR selectively with Gaussian-fidelity weight maps instead of uniformly (SRGS)3DGS + StableSR · Tanks & Temples (4x SR)FID: 49.11 -> 37.72Improved
apply SR selectively instead of uniformly (SRGS)3DGS + StableSR · Tanks & Temples (4x SR)PSNR: 23.32 -> 23.81Improved
apply SR selectively instead of uniformly (SRGS)3DGS + StableSR · Deep Blending (4x SR)PSNR: 28.23 -> 29.01Improved
apply SR selectively instead of uniformly (SRGS)3DGS + StableSR · Mip-NeRF 360 (4x SR)LPIPS: 0.317 -> 0.323Hurt
apply SR selectively instead of uniformly (SRGS)3DGS + StableSR · Mip-NeRF 360 (4x SR)PSNR: 25.92 -> 26.34Improved
use SplatSuRe SR training vs Mip-Splatting on densely sampled scenes3DGS · Mip-NeRF 360 (4x SR)PSNR: 26.48 vs 26.34Hurt
add selective SR to 3DGS trained on LR only3DGS (LR) · Tanks & Temples (4x SR)PSNR: 19.41 -> 23.81Improved
add selective SR to 3DGS trained on LR only3DGS (LR) · Tanks & Temples (4x SR)NIQE: 3.412 -> 3.928Hurt
use SwinIR instead of StableSR as SR model (ablation)SplatSuRe · Tanks & Temples (4x SR)PSNR: 23.81 -> 24.06Mixed
increase ratio threshold so more SR is applied (excessive use) (ablation)SplatSuRe · Tanks & TemplesPSNR / LPIPSMixed
SplatSuRe over SRGS with SwinIRSplatSuRe · Tanks & Temples (4x SR, SwinIR)PSNR: 23.84 -> 24.06Improved

Earlier methods this paper compares with

Named baselines from earlier accepted papers in this paper's results tables: better than 1 of 1 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