PanoPlane: Plane-Aware Panoramic Completion for Sparse-View Indoor 3D Gaussian Splatting

2026-05-13

AI Summary

This work addresses the challenges of geometric incompleteness and generation artifacts in novel view synthesis under sparse indoor viewpoints by introducing a training-free Layout Anchored Attention Steering mechanism. The method first reconstructs complete room geometry through 360° panoramic completion and then leverages detected planar structures to guide the attention of a diffusion model, enabling geometrically consistent surface extrapolation. By providing high-quality supervision signals for 3D Gaussian Splatting, the approach achieves state-of-the-art performance using only three input views on Replica, ScanNet++, and Matterport3D datasets, with PSNR improvements of up to 17.8%.

Abstract

We present PanoPlane, an approach for high-fidelity sparse-view indoor novel view synthesis that reconstructs closed room geometry via panoramic scene completion. Unlike perspective-based methods that generate training views from limited fields of view, PanoPlane leverages $360^{\circ}$ panoramic completion to condition the generative process on the full spatial layout. We propose Layout Anchored Attention Steering, a training-free mechanism that steers attention within the diffusion model's internal representation toward scene's detected planar surfaces at inference time. By directing each unobserved region's attention toward geometrically consistent observed content, our method replaces unconstrained hallucination with grounded surface extrapolation. The resulting panoramic completions provide supervision for 3D Gaussian Splatting, enabling accurate novel-view synthesis across unobserved regions from as few as three input views. Experiments on Replica, ScanNet++, and Matterport3D demonstrate state-of-the-art novel view synthesis quality across 3, 6, and 9 input views, achieving up to $+17.8\%$ improvement in PSNR over the current state-of-the-art baseline without any training or fine-tuning of the diffusion model.

Problem

• sparse-view
• indoor 3D reconstruction
• novel view synthesis
• panoramic completion
• 3D Gaussian Splatting

Innovation

• panoramic completion
• Layout Anchored Attention Steering
• 3D Gaussian Splatting
• sparse-view synthesis
• plane-aware generation