3DV 2026

PAPR Up-close:
Close-up Neural Point Rendering without Holes

Yanshu Zhang· Chirag Vashist· Shichong Peng· Ke Li

Simon Fraser University

Close-up scene DPBRF result SNP result 3DGS result PAPR result Ours result
Close-up View DPBRF SNP 3DGS PAPR Ours

Recent neural point renderers produce holes and artifacts in close-up views. Our method, PAPR Up-close, significantly reduces holes and artifacts, producing cleaner close-up renderings from a sparse point cloud.

Abstract

Point-based representations have recently gained popularity in neural rendering. While they offer many advantages, rendering them from close-up views often results in holes. In splatting-based neural point renderers, these are caused by gaps between different splats, which cause many rays to not intersect with any splat when viewed close-up. A different line of work uses attention to estimate each ray's intersection by interpolating between nearby points. Our work builds on one such method, known as Proximity Attention Point Rendering (PAPR), which learns parsimonious and geometrically accurate point representations. While in principle PAPR can fill holes by learning to interpolate between nearby points appropriately, PAPR also produces holes when rendering close-up, as the intersection point is often predicted incorrectly. We analyze this phenomenon and propose two novel solutions: a method for dynamically selecting nearby points to a ray for interpolation, and a robust attention method that better generalizes to local point configurations around unseen rays. These significantly reduce the prevalence of holes and other artifacts in close-up rendering compared to recent neural point renderers.

Method

Why Do Holes Appear?

Splatting holes

(a) Splatting-based renderers: gaps between splats cause holes

Attention holes

(b) Attention-based renderers: PAPR's cylinder selection vs. our dynamic selection (DPS)

In splatting-based methods, gaps between splats cause holes. In PAPR, holes arise because (1) the true intersection may fall outside the convex hull of selected points, and (2) the attention fails to generalize to unseen ray configurations.

Our Solutions

Dynamic Point Selection (DPS): Instead of selecting neighborhood points based solely on orthogonal distance to the ray, we use a learned weighted combination of the orthogonal distance and the distance along the ray. The weights are optimized during training to align with the attention mechanism's assessment of point relevance, dynamically adapting the local neighborhood to better capture the true intersection.

Robust Attention: We normalize the ray-dependent point features to maintain consistent scale regardless of camera distance, and introduce a Ray Perturbation Strategy (RPS) that augments ray directions during training. This exposes the attention mechanism to a more diverse range of local point configurations, improving generalization to unseen close-up views.

Ray perturbation illustration

Ray Perturbation Strategy: the ray endpoint is translated on the image plane while keeping the origin fixed.

Results

Quantitative Comparison

Method Close-up Synthetic Dataset Close-up Real Dataset
PSNR ↑ SSIM ↑ LPIPS ↓ PSNR ↑ SSIM ↑ LPIPS ↓
DPBRF13.060.3520.5257.3010.0760.779
SNP14.410.5730.45113.490.5990.608
Point-NeRF19.480.7460.39613.900.7060.551
3DGS21.450.7630.29118.150.7820.423
Mip-Splatting22.960.7860.28618.970.7940.404
PAPR18.550.7200.35716.410.7280.499
Ours 24.18 0.858 0.273 19.20 0.825 0.400

All methods use 30,000 points. Our method outperforms all baselines on all metrics.

Qualitative Comparison — Synthetic Scenes

Lego Lego DPBRF Lego SNP Lego PointNeRF Lego 3DGS Lego PAPR Lego Ours Lego GT
Hotdog Hotdog DPBRF Hotdog SNP Hotdog PointNeRF Hotdog 3DGS Hotdog PAPR Hotdog Ours Hotdog GT
Chair Chair DPBRF Chair SNP Chair PointNeRF Chair 3DGS Chair PAPR Chair Ours Chair GT
Mic Mic DPBRF Mic SNP Mic PointNeRF Mic 3DGS Mic PAPR Mic Ours Mic GT
DPBRF SNP Point-NeRF 3DGS PAPR Ours GT

Qualitative Comparison — Real Scenes

Cup Cup DPBRF Cup SNP Cup PointNeRF Cup 3DGS Cup PAPR Cup Ours Cup GT
Bowser Bowser DPBRF Bowser SNP Bowser PointNeRF Bowser 3DGS Bowser PAPR Bowser Ours Bowser GT
DPBRF SNP Point-NeRF 3DGS PAPR Ours GT

Ablation Study

We incrementally add our proposed components to the vanilla PAPR baseline.

PAPR baseline + Robust Attention + DPS (Full) Ground Truth
PAPR + Robust Attn + DPS (Full) Ground Truth

Robust attention significantly improves hole-filling. Adding DPS further fills remaining gaps and refines surface geometry.

Citation

@inproceedings{zhang2026paprupclose,
  title     = {PAPR Up-close: Close-up Neural Point
               Rendering without Holes},
  author    = {Zhang, Yanshu and Vashist, Chirag
               and Peng, Shichong and Li, Ke},
  booktitle = {International Conference on
               3D Vision (3DV)},
  year      = {2026}
}