OpenQuestCapture for Quest 3 and 3S Capture

Michael Rubloff

Michael Rubloff

OpenQuestCapture

OpenQuestCapture is a new open source project that pushes the Meta Quest 3 and 3S closer to a surprisingly potent capture device. Built as a Unity application and released under an MIT license, the repository allows users to directly capture from the headset.

With the surge in popularity of Meta’s Hyperscape Capture, there has been a growing awareness, and demand, for capturing directly with a headset. Unfortunately, Meta does not currently offer a way to export either raw capture data or trained PLY files. What it does allow is the creation of shareable links, enabling up to eight people to join a space simultaneously, but the underlying data remains locked away.

OpenQuestCapture fills that gap by capturing headset and controller poses, stereo passthrough imagery, camera intrinsics, and raw depth maps, all aligned in time and organized into session based logs. The project exposes the full capture pipeline, from Camera2 API characteristics to raw YUV frames and per camera depth descriptors.

Similar to Hyperscape Capture, OpenQuestCapture also gives real time visualization inside the headset. As users scan an environment, they can see coverage feedback rendered live, with color indicating the angle of capture. White for head on views and increasingly vivid colors for grazing angles.

The project is also paired with a companion repository, Meta Quest 3D Reconstruction, which handles parsing, decoding, and reconstruction of the captured data. Together, the two projects form a complete pipeline that converts raw Quest logs into COLMAP compatible outputs using Open3D and Python utilities. From there, scenes can be passed directly into Gaussian Splatting workflows, like Postshot or Lichfeld Studio.

OpenQuestCapture is built on top of the Quest sensor data capture foundations laid by QuestRealityCapture, created by t-34400. The author explicitly credits that work for making Quest sensor access feasible in the first place. The app is sideloaded onto the Quest, and you will need to enable permissions for headset cameras, and recording is toggled using the left controller’s menu button.

Sessions are automatically stored in timestamped folders on internal storage, with options to export recordings as zip files directly from within the headset. For those who want a fully managed path, captured data can be uploaded to vid2scene’s cloud service, which handles reconstruction and produces a 3D Gaussian Splatting scene without local setup.

The project is still early, with only a handful of contributors and modest GitHub traction so far, but its the latest foray into capturing directly with a Quest. Just a few months ago, it was hard to imagine anyone using their headset to be capturing gaussian splatting data with, but it’s being rapidly proven that the headset form factor is a compelling capture device.

Learn more and access OpenQuestCapture through their GitHub repo!

Featured

Recents

Platforms

CHCNAV CoMeta Exports 3D Gaussian Splats to 3D Tiles

CHCNAV CoMeta generates 3D Gaussian splatting from the same aerial triangulation as its orthomosaic, DSM and mesh, and exports splats to PLY and 3D Tiles.

Michael Rubloff

Sep 14, 2026

Platforms

CHCNAV CoMeta Exports 3D Gaussian Splats to 3D Tiles

CHCNAV CoMeta generates 3D Gaussian splatting from the same aerial triangulation as its orthomosaic, DSM and mesh, and exports splats to PLY and 3D Tiles.

Michael Rubloff

Research

Gaussian Light Transport Fits Global Illumination as a Mixture of 13D Gaussians

aussian Light Transport fits global illumination as a mixture of 13D Gaussians, training 10 to 23 times faster than Neural Radiosity in a few megabytes.

Michael Rubloff

Sep 14, 2026

Research

Gaussian Light Transport Fits Global Illumination as a Mixture of 13D Gaussians

aussian Light Transport fits global illumination as a mixture of 13D Gaussians, training 10 to 23 times faster than Neural Radiosity in a few megabytes.

Michael Rubloff

Platforms

Spark v2.2.0 Adds Faster LoD Traversal, Rust Decoders

v2.2.0 adds an experimental faster LoD splat traversal algorithm, replaces JavaScript splat decoders with Rust ones, and reports up to 20% faster sorting.

Michael Rubloff

Sep 14, 2026

Platforms

Spark v2.2.0 Adds Faster LoD Traversal, Rust Decoders

v2.2.0 adds an experimental faster LoD splat traversal algorithm, replaces JavaScript splat decoders with Rust ones, and reports up to 20% faster sorting.

Michael Rubloff

Platforms

ANNX Studio v1.1.11 Adds 360 Photo Project Creation

ANNX Studio v1.1.11 rebuilds its 360 panorama modeling pipeline as 2.0 and adds project creation from 360 still images, previously limited to 360 video.

Michael Rubloff

Sep 11, 2026

Platforms

ANNX Studio v1.1.11 Adds 360 Photo Project Creation

ANNX Studio v1.1.11 rebuilds its 360 panorama modeling pipeline as 2.0 and adds project creation from 360 still images, previously limited to 360 video.

Michael Rubloff

Trending Articles
  1. TRENDING
    Loading...
  2. TRENDING
    Loading...
  3. TRENDING
    Loading...
Michael Rubloff

Written by Michael Rubloff

Michael is the Founder and Managing Editor of Radiancefields.com

More from Michael Rubloff

More from Michael Rubloff

Platforms

CHCNAV CoMeta Exports 3D Gaussian Splats to 3D Tiles

CHCNAV CoMeta generates 3D Gaussian splatting from the same aerial triangulation as its orthomosaic, DSM and mesh, and exports splats to PLY and 3D Tiles.

Michael Rubloff

Sep 14, 2026

CHCNAV

Platforms

CHCNAV CoMeta Exports 3D Gaussian Splats to 3D Tiles

Michael Rubloff

Sep 14, 2026

CHCNAV

Platforms

CHCNAV CoMeta Exports 3D Gaussian Splats to 3D Tiles

CHCNAV CoMeta generates 3D Gaussian splatting from the same aerial triangulation as its orthomosaic, DSM and mesh, and exports splats to PLY and 3D Tiles.

Michael Rubloff

Sep 14, 2026

CHCNAV

Research

Gaussian Light Transport Fits Global Illumination as a Mixture of 13D Gaussians

aussian Light Transport fits global illumination as a mixture of 13D Gaussians, training 10 to 23 times faster than Neural Radiosity in a few megabytes.

Michael Rubloff

Sep 14, 2026

Gaussian Light Transport

Research

Gaussian Light Transport Fits Global Illumination as a Mixture of 13D Gaussians

Michael Rubloff

Sep 14, 2026

Gaussian Light Transport

Research

Gaussian Light Transport Fits Global Illumination as a Mixture of 13D Gaussians

aussian Light Transport fits global illumination as a mixture of 13D Gaussians, training 10 to 23 times faster than Neural Radiosity in a few megabytes.

Michael Rubloff

Sep 14, 2026

Gaussian Light Transport

Platforms

Spark v2.2.0 Adds Faster LoD Traversal, Rust Decoders

v2.2.0 adds an experimental faster LoD splat traversal algorithm, replaces JavaScript splat decoders with Rust ones, and reports up to 20% faster sorting.

Michael Rubloff

Sep 14, 2026

Spark

Platforms

Spark v2.2.0 Adds Faster LoD Traversal, Rust Decoders

Michael Rubloff

Sep 14, 2026

Spark

Platforms

Spark v2.2.0 Adds Faster LoD Traversal, Rust Decoders

v2.2.0 adds an experimental faster LoD splat traversal algorithm, replaces JavaScript splat decoders with Rust ones, and reports up to 20% faster sorting.

Michael Rubloff

Sep 14, 2026

Spark