AI Paper Atlas

Research map / 3D vision / Gaussian splatting

PATexGS: Perceptual-Adaptive Texture Scheduling for Visual Coherence in Textured Gaussian Splatting

Yuesong Wang, Dounian Ma, Xiaoyu Chen, Tao Guan

AAAI 2026 · 3D vision › Gaussian splatting · approach: Regularization and primitive design for 3DGS · 0 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 mipmap-based texture scheduling at 1/8 resolution (ablation)PATexGS · Tanks&Temples 10% Gaussians, 1/8 resolutionPSNR: 21.15 -> 23.65Improved
add mipmap scheduling at full resolution (ablation)PATexGS · Tanks&Temples 10% Gaussians, full resolutionPSNR: 23.41 -> 23.43No clear effect
add alpha-channel mipmap (ablation)PATexGS · Tanks&Temples 10% Gaussians, 1/8 resolutionPSNR: 21.59 -> 23.65Improved
add RGB-channel mipmap (ablation)PATexGS · Tanks&Temples 10% Gaussians, 1/8 resolutionPSNR: 23.35 -> 23.65Improved
use adaptive texture allocation instead of fixed 32x32 textures (ablation)PATexGS · Tanks&Temples 10% GaussiansPSNR: 23.02 -> 23.43Improved
use adaptive allocation instead of fixed 32x32 textures (storage) (ablation)PATexGS · Tanks&Temples 10% Gaussiansstorage (M): 223.0 -> 45Improved
use adaptive allocation instead of fixed texture of similar parameter count (ablation)PATexGS · Tanks&Temples 10% GaussiansPSNR: 23.29 -> 23.43Improved
apply adaptive allocation to BBSplatBBSplat · Mip-NeRF360 (9 scenes), full GaussiansPSNR: 26.92 -> 27.11 (Mip-NeRF360); 23.69 -> 24.10 (T&T); SSIM 0.786 -> 0.785Mixed
use PATexGS instead of 2DGS when only 1% of Gaussians are kept2DGS · Mip-NeRF360 (9 scenes), 1% GaussiansPSNR: 22.64 -> 23.84Improved
use PATexGS instead of Mip Splatting at full GaussiansMip Splatting · Mip-NeRF360 (9 scenes), full GaussiansPSNR: 27.79 -> 27.71Hurt

Earlier methods this paper compares with

Named baselines from earlier accepted papers in this paper's results tables: better than 0 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