AI Paper Atlas

Research map / 3D vision / Gaussian splatting

Z-Order Transformer for Feed-Forward Gaussian Splatting

Can Wang, Lei Liu, Wei Jiang, Dong Xu

CVPR 2026 · 3D vision › Gaussian splatting · approach: Regularization and primitive design for 3DGS · 1 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 Z-order based ZFormer blocks (vs convolutional layers) (ablation)Z-order transformer · RealEstate10K 360x640, 12 viewsPSNR: 24.86 -> 28.56Improved
add sparse attention (vs full attention) (ablation)Z-order transformer · RealEstate10K 360x640, 12 viewsPSNR: 26.79 -> 28.56Improved
initialize SH parameters from pixel color (ablation)Z-order transformer · RealEstate10K 360x640, 12 viewsPSNR: 27.81 -> 28.56Improved
train depth estimator jointly instead of fixing it (ablation)Z-order transformer · RealEstate10K 360x640, 12 viewsPSNR: 28.15 -> 28.56Improved
add top-k selection attention on top of group attention (ablation)sparse attention · RealEstate10K 360x640, 12 viewsPSNR: 26.79 -> 28.56Improved
Z-order view selection vs random selection (24 views to 16)Z-order transformer · RealEstate10K 24 input viewsPSNR: 27.97 -> 28.73Improved
Z-order view selection instead of using all 24 viewsZ-order transformer · RealEstate10K 24 input viewsPSNR: 28.91 -> 28.73 (time 0.622s -> 0.448s)Hurt
Z-order view selection instead of random (16 to 8 views)Z-order transformer · RealEstate10K 16 input viewsPSNR: 27.06 -> 28.07Improved
second Z-order block (L2 vs L1) to compress GaussiansZ-order transformer · RealEstate10K 2 viewsnumber of Gaussians (x10^5): 2.85 -> 1.42Improved
second Z-order block (L2 vs L1)Z-order transformer · RealEstate10K 360x640, 12 viewsPSNR: 28.56 -> 28.12Hurt
use pretrained VGGT backbone instead of depth-anything-v2-smallZ-order transformer · RealEstate10K 360x640, 12 viewsPSNR: 28.56 -> 28.81 (time 0.337s -> 0.815s)Mixed

Earlier methods this paper compares with

Named baselines from earlier accepted papers in this paper's results tables: better than 2 of 2 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