AI Paper Atlas

Research map / 3D vision / Gaussian splatting

Horseshoe Splatting: Handling Structural Sparsity for Uncertainty-Aware Gaussian-Splatting Radiance Field Rendering

Feng Wu, Tsai Hor Chan, Yihang Chen, Lingting Zhu, Guosheng Yin, Lequan Yu

ICLR 2026 · 3D vision › Gaussian splatting · approach: Compact and efficient Gaussian splatting

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 Horseshoe prior instead of Gaussian prior on Gaussian scales (ablation)Horseshoe Splatting · LFNLL: 9.15 -> -0.74Improved
add Horseshoe prior instead of Laplace prior on Gaussian scales (ablation)Horseshoe Splatting · LFNLL: 10.58 -> -0.74Improved
add Horseshoe prior instead of Laplace prior (ablation)Horseshoe Splatting · LLFFAUSE: 0.42 -> 0.31Improved
add Horseshoe prior instead of Gaussian/Laplace prior (ablation)Horseshoe Splatting · LFPSNR: 30.01 -> 30.05No clear effect
add Horseshoe prior instead of Laplace prior (ablation)Horseshoe Splatting · LLFFNLL: 8.22 -> 0.14Improved
use Horseshoe prior instead of Laplace for sparsity of scales (ablation)Horseshoe Splatting · LF torch sparsitysparsity ratio: 61.01 -> 64.62Improved
increase log-scale prior variance rho from -5 to -1 (ablation)Horseshoe Splatting · LLFFPSNR: 25.86 -> 25.49Hurt
raise rho from -5 to -2 (ablation)Horseshoe Splatting · LLFFPSNR: 25.86 -> 25.96Mixed
lower KL weight from 0.1 to 0.001 (ablation)Horseshoe Splatting · LLFFPSNR: 25.61 -> 25.86Improved
lower KL weight further from 0.001 to 0.0001 (ablation)Horseshoe Splatting · LLFFNLL: 0.14 -> 0.16Hurt
use uncertainty-guided active view selection instead of randomHorseshoe Splatting · LLFF active learningPSNR: 20.97 -> 26.23Improved
use Horseshoe Splatting instead of Variational 3DGSVariational 3DGS · Tanks & TemplesNLL: 2.46 -> 0.58Improved

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