Open4D

Open4D is an open, research-driven platform for the representation, compression, processing, evaluation, and streaming of time-varying 4D geometry data. It is designed as shared infrastructure for application domains such as XR systems, robotics, teleoperation, digital twins, and autonomous systems, where geometry evolves and varies over time and must be processed under tight latency, bandwidth, and accuracy constraints.

---

Ecosystem Overview

Open4D ecosystem layers:

• Core (open4d/): canonical 4D data models, IO, metrics
• Modules (modules/): research algorithms (e.g., TVMC, ARAP-based tracking)
• Benchmarks (benchmarks/): paper-reproducible evaluations
• Apps (apps/): end-to-end pipelines and demos
• Bindings (cpp/, python/): performance-critical + research-friendly APIs

---

Why Open4D?

Today, 4D geometry pipelines are fragmented:

• algorithms live in paper-specific codebases,
• evaluation scripts are ad hoc,
• datasets and metrics are inconsistent,
• systems integration is repeatedly re-implemented.

Open4D provides a common substrate:

• stable 4D data abstractions,
• reusable IO and metrics,
• modular research algorithms,
• reproducible benchmarks,
• end-to-end system pipelines.

The goal is to let research compound, not reset with every paper.

---

Repository Structure

Open4D/
├── open4d/ # Core public API (stable)
│ ├── core/ # 4D data structures
│ ├── io/ # Readers / writers
│ ├── metrics/ # Quality + temporal metrics
│ └── utils/
├── modules/ # Research systems and algorithms
│ ├── tvmc/
│ ├── arap_volume_tracking/
│ └── tvm_editing/
├── benchmarks/ # Reproducible experiments
├── apps/ # End-to-end pipelines
├── examples/ # Minimal usage examples
├── cpp/ # C++ core + bindings
├── tests/ # Unit + integration tests
├── docs/ # Documentation and figures
└── docker/ # Reproducible environments

---

Design Principles

• 4D-first: time is a first-class signal, not an afterthought
• Separation of concerns: core abstractions vs. research modules
• Reproducibility: benchmarks are explicit and scriptable
• Systems-aware: metrics include bitrate, latency, and temporal stability
• Cross-domain: XR, robotics, and autonomy share the same foundations

---

Current Modules

• TVMC — Time-Varying Mesh Compression
• Unity Decoder Plugin - a 4D mesh decoder plugin for Unity that can run on XR headsets like Meta Quest3
• ARAP Volume Tracking — deformation-aware temporal alignment
• Temporal Mesh Editing — structured editing of dynamic geometry

(Additional modules are expected to evolve independently on top of the core API.)

---

Installation (Development)

git clone https://github.com/open4d/open4d.git
cd Open4D
pip install -e .

Citation

If you use Open4D in academic work, please cite the project:

@software{open4d,
  title  = {Open4D: An Open-Source Library for 4D Geometry Processing},
  author = {Chen, Guodong and Dasari, Mallesham},
  year   = {2025}
} 

Contributors

Core contributors:

• Jason Balayev (balayev.j@northeastern.edu)
• Guodong Chen (chen.guod@northeastern.edu)
• Mallesham Dasari (m.dasari@northeastern.edu)

Contributions from the broader research and open-source community are welcome.

Contributing

Open4D is an open research platform. Contributions are welcome in the form of:

• new modules
• benchmarks and datasets
• metrics
• documentation
• performance improvements

Please reach out if you'd like to contribute!