Gallery

Sample Figures

Architecture diagrams, pipeline overviews, and evaluation-style plots from the Open4D platform. Charts marked illustrative show the shape of Open4D evaluations, not measured results — real benchmark numbers live on the Benchmarks page as they are published.

Open4D ecosystem architecture diagram showing core, modules, benchmarks, apps, and bindings layers — **Open4D ecosystem.** The five layers of the platform — core API, research modules, benchmarks, apps, and bindings — from the project documentation.

End-to-end 4D pipeline: capture, encode, stream, decode, render — **End-to-end 4D pipeline.** Open4D covers the full path from capture to on-headset rendering: encoding with N4MC/TSMC/TVMC, streaming under bitrate and latency budgets, and decoding via the Unity plugin or the Python player.

A time-varying mesh sequence with changing vertex counts across four frames — **Time-varying meshes.** Unlike animated meshes with fixed connectivity, time-varying meshes change topology and vertex count every frame — the core representation challenge Open4D's data model and codecs are designed for.

Illustrative rate-distortion curves comparing N4MC, TVMC, and a per-frame baseline — **Rate–distortion evaluation.** The standard Open4D benchmark view: geometry quality versus bitrate for temporal codecs against per-frame baselines. *Illustrative curves.*

Illustrative bar chart of per-frame decode latency for N4MC, TSMC, TVMC, and a per-frame baseline — **Systems-aware metrics.** Decode latency per frame is a first-class metric — real-time XR playback needs the whole decode–render loop inside a frame budget. *Illustrative values.*

📷

Have renders, demo videos, or results produced with Open4D? We would love to feature them here — get in touch or open a pull request against the repository.