2DGS-R: Revisiting the Normal Consistency Regularization in 2D Gaussian Splatting

2025-10-19

AI Summary

To address the trade-off between geometric fidelity and rendering quality in 2D Gaussian Splatting (2DGS), this paper proposes a hierarchical training framework that decouples geometry reconstruction from appearance modeling. Methodologically, it introduces normal consistency regularization to enforce surface geometric plausibility, designs a native Gaussian cloning mechanism for structure-aware adaptive refinement, and adopts a transparency-freezing fine-tuning strategy to stabilize optimization. Built upon the 2D Gaussian disk representation, the framework achieves significant improvements in rendering metrics (e.g., +1.2 dB PSNR, +0.015 SSIM) with negligible overhead—only a +1% increase in storage and no additional training cost. Crucially, it maintains high-fidelity implicit geometry reconstruction, with signed distance function (SDF) error consistently below 0.15 mm. The core contribution is the first unified framework enabling both high-quality novel-view synthesis and precise implicit geometric representation within a single 2DGS architecture.

Technology Category

• Computer Vision: Diffusion Models for Vision
• Knowledge Representation and Reasoning: Geometric, Spatial, and Temporal Reasoning
• Machine Learning: Representation Learning

Application Category

• Security and Privacy: Data transparency and provenance
• Economics, Online Markets and Human Computation: Data quality aspects of human-annotated datasets
• Graph Algorithms and Modeling for the Web: Efficient manipulation of static and dynamic Web-related graphs

Abstract

Recent advancements in 3D Gaussian Splatting (3DGS) have greatly influenced neural fields, as it enables high-fidelity rendering with impressive visual quality. However, 3DGS has difficulty accurately representing surfaces. In contrast, 2DGS transforms the 3D volume into a collection of 2D planar Gaussian disks. Despite advancements in geometric fidelity, rendering quality remains compromised, highlighting the challenge of achieving both high-quality rendering and precise geometric structures. This indicates that optimizing both geometric and rendering quality in a single training stage is currently unfeasible. To overcome this limitation, we present 2DGS-R, a new method that uses a hierarchical training approach to improve rendering quality while maintaining geometric accuracy. 2DGS-R first trains the original 2D Gaussians with the normal consistency regularization. Then 2DGS-R selects the 2D Gaussians with inadequate rendering quality and applies a novel in-place cloning operation to enhance the 2D Gaussians. Finally, we fine-tune the 2DGS-R model with opacity frozen. Experimental results show that compared to the original 2DGS, our method requires only 1% more storage and minimal additional training time. Despite this negligible overhead, it achieves high-quality rendering results while preserving fine geometric structures. These findings indicate that our approach effectively balances efficiency with performance, leading to improvements in both visual fidelity and geometric reconstruction accuracy.

Problem

• Improving rendering quality while maintaining geometric accuracy
• Overcoming limitations in 2D Gaussian Splatting surface representation
• Balancing efficiency with performance in neural field reconstruction

Innovation

• Hierarchical training approach for rendering quality
• In-place cloning operation to enhance 2D Gaussians
• Fine-tuning with opacity frozen for geometric accuracy