Skip to main content
GameDev.net gamedev.net

Point Clouds News

The latest Point Clouds coverage curated for game developers.

PointGrade: Geometric Priors for Grading MoonBoard Problems

arXiv cs.GR details PointGrade, a machine-learning method that grades MoonBoard bouldering problems by sampling point clouds from pre-scanned hold meshes. By folding geometric priors into a sequence-based predictor, the model …

Research arXiv cs.GR · 6 days, 9 hours ago
Lipschitz Extension Initialization for Moving Least Squares Reconstruction from Sparse Irregular Samples

A new reconstruction method uses a Lipschitz-extension pass to stabilize Moving Least Squares on sparse, irregular samples. The result is better accuracy and fewer failures when data points are unevenly …

Research arXiv cs.GR · 2 weeks, 5 days ago
PointGT: Simultaneous Geometry and Texture Editing for Point-Based Representations

PointGT is a new point-based 3D representation that lets teams edit geometry and texture at the same time. The method targets a long-standing gap in neural reconstruction: photorealistic assets are …

Research arXiv cs.GR · 2 weeks, 5 days ago
P-CORE: Self-Supervised Surface Consistency for Point-Based Neural Editing

A new self-supervised method called P-CORE aims to make point-based neural editing survive large non-rigid deformations without ground-truth views of the edited shape. It enforces surface consistency before and after …

Research arXiv cs.GR · 2 weeks, 5 days ago
Unified Motion Retargeting for Humanoids with Learned Point Cloud Correspondence

A new humanoid retargeting system uses learned point-cloud correspondence to map human motion onto robots without hand-built keypoint pairs. That matters for teams training embodied agents: it promises denser pose …

Research arXiv cs.GR · 2 weeks, 6 days ago
MVFM-3DAD: Multi-view Flow Matching for 3D Anomaly Detection via Density Proxy Estimation

A new 3D anomaly detection method, MVFM-3DAD, uses multi-view flow matching to score point clouds by how well they fit a learned normal distribution. It avoids both memory-bank coverage gaps …

Research arXiv cs.GR · 1 month, 1 week ago
Learning Manifolds in High-D Point Embedding for Anisotropic Surface Approximation from Unstructured Point Clouds

A new graphics paper proposes HD-PEA, a learning-based pipeline for turning unstructured point clouds into compact anisotropic surface approximations. The method embeds points into a high-dimensional manifold, then estimates tangent …

Research arXiv cs.GR · 1 month, 2 weeks ago
Points as Tori: Fast Pointwise Signed Distance for Point Clouds

A new point-cloud distance method turns each sample into a locally fitted torus, giving fast signed-distance queries without global optimization or voxelization. For graphics teams, that means point clouds can …

Research arXiv cs.GR · 2 months ago
IBPA: Real-time Free-form Manifold Mesh Reconstruction via Incremental Ball Pivoting with Integrated Hole Detection

A new reconstruction method turns streaming point-cloud data into an orientable manifold mesh in real time, aiming to help underwater teams spot coverage gaps while they are still on site. …

Research arXiv cs.GR · 2 months, 1 week ago
Benchmarking Federated Learning and Knowledge Distillation for Point Cloud Classification

Federated learning for 3D point-cloud classification is looking far less plug-and-play than it sounds. A 504-run benchmark across 13 FL methods and 10 distillation objectives found that extreme non-IID label …

Research arXiv cs.GR · 2 months, 2 weeks ago
Skeletal-Anchored Dual Harmonics for Structured 3D Modeling

A new 3D shape representation uses internal skeletal anchors plus dual-channel spherical harmonics to capture both surface detail and structural organization. For graphics teams, the interesting part is that it …

Research arXiv cs.GR · 3 months, 2 weeks ago
RigPAPR: Rig-Based Animation of Static Neural Point Clouds from a Fixed-Viewpoint Video

RigPAPR tackles a common failure mode in neural character animation: joint-boundary artifacts when Gaussian splats are driven by rigid skinning. The paper claims a static PAPR point cloud can be …

Research arXiv cs.GR · 3 months, 2 weeks ago
GS-NFS: Bandwidth-adaptive Streaming of Dynamic Gaussian Splats and Point Clouds

GPU-side compression/decompression for dynamic Gaussian splats is the real story here: GS-NFS claims full-frame-rate streaming with 1-2 orders of magnitude better encode/decode speed than prior work. For teams watching 3D …

Research arXiv cs.GR · 3 months, 2 weeks ago
Composable function systems as a general-purpose rendering framework

A new arXiv paper argues function systems can do more than fractals: they can serve as a meshless rendering and simulation framework with very low GPU/CPU memory use. The author …

Research arXiv cs.GR · 3 months, 3 weeks ago
BrickAnything: Geometry-Conditioned Buildable Brick Generation with Structure-Aware Tokenization

A new AI pipeline can turn point clouds into physically buildable brick assemblies instead of just plausible-looking shapes. BrickAnything adds structure-aware tree tokenization plus validity checks, rollback, and preference-based alignment …

Research arXiv cs.GR · 3 months, 3 weeks ago
Curve Skeletonization in Continuous domain for Meshes and Point Clouds

This paper pushes curve skeletonization out of the usual discrete graph setup and into a continuous domain, which matters if you care about cleaner topology from messy meshes or point …

Research arXiv cs.GR · 3 months, 4 weeks ago
PointLLM-R: Enhancing 3D Point Cloud Reasoning via Chain-of-Thought

A 55K-sample dataset is the real headline here: the authors use chain-of-thought supervision to make 3D point-cloud models reason instead of just classify. For teams working on scan-heavy games, tools, …

Research arXiv cs.GR · 4 months ago
PolycubeNet: A Dual-latent Diffusion Model for Polycube-Based Hexahedral Mesh Generation

A new diffusion-based pipeline can generate polycube representations for hexahedral meshing directly from point clouds, skipping the usual manual segmentation and template-heavy setup. For teams doing simulation-heavy work, the interesting …

Research arXiv cs.GR · 4 months ago
QuadLink: Autoregressive Quad-Dominant Mesh Generation via Point-Relation Learning

QuadLink tackles a real production pain point for 3D teams: turning point clouds into usable quad-dominant meshes with coherent edge flow. The paper claims better geometric fidelity and topology than …

Research arXiv cs.GR · 4 months ago
StippleDiffusion: Capacity-Constrained Stippling using Controlled Diffusion

A diffusion model now does capacity-constrained stippling in one pass, instead of relying on slow iterative optimizers that must be rerun for every target image. The paper says the sampler …

Research arXiv cs.GR · 4 months ago