# Resolution Where It Counts: Hash-based GPU-Accelerated 3D Reconstruction via Variance-Adaptive Voxel Grids

arXiv: https://arxiv.org/abs/2511.21459 · HTML: https://mlsanity.com/paper/2511.21459/

## Key facts

- Cited by 1 papers in the map (data through 14 Oct 2026).
- Builds on 24 of its 45 references that are in the map.
- First posted on arXiv 26 Nov 2025.

## Builds on (most cited first, 24 of 24)

- [The Unreasonable Effectiveness of Deep Features as a Perceptual Metric](https://mlsanity.com/paper/1801.03924/) (2018)
- [3D Gaussian Splatting for Real-Time Radiance Field Rendering](https://mlsanity.com/paper/2308.04079/) (2023)
- [Instant Neural Graphics Primitives with a Multiresolution Hash Encoding](https://mlsanity.com/paper/2201.05989/) (2022)
- [ScanNet: Richly-annotated 3D Reconstructions of Indoor Scenes](https://mlsanity.com/paper/1702.04405/) (2017)
- [Occupancy Networks: Learning 3D Reconstruction in Function Space](https://mlsanity.com/paper/1812.03828/) (2018)
- [The Replica Dataset: A Digital Replica of Indoor Spaces](https://mlsanity.com/paper/1906.05797/) (2019)
- [SplaTAM: Splat, Track & Map 3D Gaussians for Dense RGB-D SLAM](https://mlsanity.com/paper/2312.02126/) (2023)
- [NICER-SLAM: Neural Implicit Scene Encoding for RGB SLAM](https://mlsanity.com/paper/2302.03594/) (2023)
- [Voxblox: Incremental 3D Euclidean Signed Distance Fields for On-Board MAV Planning](https://mlsanity.com/paper/1611.03631/) (2016)
- [BundleFusion: Real-time Globally Consistent 3D Reconstruction using On-the-fly Surface Re-integration](https://mlsanity.com/paper/1604.01093/) (2016)
- [Neural Kernel Surface Reconstruction](https://mlsanity.com/paper/2305.19590/) (2023)
- [The Oxford Spires Dataset: Benchmarking Large-Scale LiDAR-Visual Localisation, Reconstruction and Radiance Field Methods](https://mlsanity.com/paper/2411.10546/) (2024)
- [GSFusion: Online RGB-D Mapping Where Gaussian Splatting Meets TSDF Fusion](https://mlsanity.com/paper/2408.12677/) (2024)
- [fVDB: A Deep-Learning Framework for Sparse, Large-Scale, and High-Performance Spatial Intelligence](https://mlsanity.com/paper/2407.01781/) (2024)
- [Multi-Camera LiDAR Inertial Extension to the Newer College Dataset](https://mlsanity.com/paper/2112.08854/) (2021)
- [VBR: A Vision Benchmark in Rome](https://mlsanity.com/paper/2404.11322/) (2024)
- [Multi-Resolution 3D Mapping with Explicit Free Space Representation for Fast and Accurate Mobile Robot Motion Planning](https://mlsanity.com/paper/2010.07929/) (2020)
- [MD-SLAM: Multi-cue Direct SLAM](https://mlsanity.com/paper/2203.13237/) (2022)
- [Splat-LOAM: Gaussian Splatting LiDAR Odometry and Mapping](https://mlsanity.com/paper/2503.17491/) (2025)
- [N3-Mapping: Normal Guided Neural Non-Projective Signed Distance Fields for Large-scale 3D Mapping](https://mlsanity.com/paper/2401.03412/) (2024)
- [MAP-ADAPT: Real-Time Quality-Adaptive Semantic 3D Maps](https://mlsanity.com/paper/2406.05849/) (2024)
- [MAD-ICP: It Is All About Matching Data -- Robust and Informed LiDAR Odometry](https://mlsanity.com/paper/2405.05828/) (2024)
- [Photometric LiDAR and RGB-D Bundle Adjustment](https://mlsanity.com/paper/2303.16878/) (2023)
- [MAD-BA: 3D LiDAR Bundle Adjustment -- from Uncertainty Modelling to Structure Optimization](https://mlsanity.com/paper/2501.03972/) (2025)

## Built on by (most cited first, 1 of 1)

- [OctWorld: Long-Range World-Consistent Video Generation with Octree-Based 3D Mapping](https://mlsanity.com/paper/2609.03919/) (2026)

## Abstract (arXiv)

Efficient and scalable 3D surface reconstruction from range data remains a core challenge in computer graphics and vision, particularly in real-time and resource-constrained scenarios. Traditional volumetric methods based on fixed-resolution voxel grids or hierarchical structures like octrees often suffer from memory inefficiency, computational overhead, and a lack of GPU support. We propose a novel variance-adaptive, multi-resolution voxel grid that dynamically adjusts voxel size based on the local variance of signed distance field (SDF) observations. Unlike prior multi-resolution approaches that rely on recursive octree structures, our method leverages a flat spatial hash table to store all voxel blocks, supporting constant-time access and full GPU parallelism. This design enables high memory efficiency and real-time scalability. We further demonstrate how our representation supports GPU-accelerated rendering through a parallel quad-tree structure for Gaussian Splatting, enabling effective control over splat density. Our open-source CUDA/C++ implementation achieves up to 13x speedup and 4x lower memory usage compared to fixed-resolution baselines, while maintaining on par results in terms of reconstruction accuracy, offering a practical and extensible solution for high-performance 3D reconstruction.