PPSURF: Combining Patches and Point Convolutions for Detailed Surface Reconstruction

TL;DR

The paper tackles robust surface reconstruction from unoriented point clouds by combining a global prior based on point convolutions with a local prior built from dense local patches, forming a two-branch PPSurf architecture. This approach preserves fine surface details while remaining resilient to noise and sampling artifacts, outperforming both non-data-driven and data-driven baselines across diverse datasets. Extensive ablations confirm the value of jointly modeling global structure and local detail, with optimal local patch sizes around $25$–$100$NN and attention-based interpolation enhancing performance. While effective, PPSurf remains non-generative for large missing regions and incurs non-interactive reconstruction times, suggesting future work to integrate generative components and faster sampling strategies.

Abstract

3D surface reconstruction from point clouds is a key step in areas such as content creation, archaeology, digital cultural heritage, and engineering. Current approaches either try to optimize a non-data-driven surface representation to fit the points, or learn a data-driven prior over the distribution of commonly occurring surfaces and how they correlate with potentially noisy point clouds. Data-driven methods enable robust handling of noise and typically either focus on a global or a local prior, which trade-off between robustness to noise on the global end and surface detail preservation on the local end. We propose PPSurf as a method that combines a global prior based on point convolutions and a local prior based on processing local point cloud patches. We show that this approach is robust to noise while recovering surface details more accurately than the current state-of-the-art. Our source code, pre-trained model and dataset are available at: https://github.com/cg-tuwien/ppsurf

Figures (5)

• Figure 2: Point cloud examples of the data sets used in our evaluation.
• Figure 3: Qualitative comparison to all baselines. We evaluate one example from each dataset variant (except for the no-noise variants, where we only show one example due to space constraints). Colors show the distance of the reconstructed surface to the ground-truth surface. Due to our combined local and global branches, PPSurf reconstructs details more accurately than the baselines, especially in the presence of strong input noise. Note that results for Neural IMLS are not provided by the authors for the high-noise dataset variants. See the supplementary material for a qualitative comparison on all shapes in our test sets.
• Figure 4: Real-world reconstructions. We compare to all baselines on the two point clouds that were obtained from real-world objects.
• Figure 5: Limitations. Our method has difficulties to recover the edges of clean point clouds due to training with noisy point clouds.
• Figure 6: Limitations. Our method struggles with reconstructions of large missing areas in the input point cloud since we did not incorporate any generative model capabilities.