splaTV: Dynamic Gaussian Splatting Viewer

Michael Rubloff

Michael Rubloff

Mar 15, 2024

Email
Copy Link
Twitter
Linkedin
Reddit
Whatsapp
SplaTV
SplaTV

Kevin Kwok, perhaps better known as Antimatter15, has released something amazing: splaTV.

It's an implementation of Spacetime Gaussians, a state-of-the-art method for dynamic Gaussian Splatting, running natively in browsers. Antimatter15's viewer for static Gaussian Splatting quickly became a go-to choice for those interested in exploring their splats, and splaTV is poised to achieve similar status for dynamic Gaussian Splatting.

Although the complexity of Spacetime Gaussians and their numerous parameters make the file format challenging to port, this approach benefits splaTV by facilitating easy loading and streaming without additional processing.

This could be a contributing factor to its ability to run at a full 120 fps in-browser. Here are the movement controls for the viewer:

  • Vertical Movement: Use the W key to move upwards and the S key to move downwards.

  • Camera Swing: To swing the camera left, press A. For right, press D.

  • Dolly Zoom: Use the square brackets [ ] keys to enable a dolly zoom effect, adding a cinematic feel to your exploration.

  • Zoom: Regular zoom is done by pinching on a Mac trackpad or doing ctrl+scroll to zoom in and out.

splaTV is additionally MIT Licensed, similar to Antimatter15's other work, allowing for permissible usage. There is also the ability to add custom scenes into the viewer. With that in mind, here are two additional sample scenes.

Scene 1 and Scene 2.

Enhancing its appeal, splaTV also supports Webgl and WebXR, enabling access in XR environments. Antimatter15 ensures users can easily immerse themselves in the scene, provided they have enabled the necessary WebXR flags in their browsers.

We have seen a few papers recently looking at dynamic splats, in addition to the announcement of Gmix, a dynamic gaussian splatting company.

Experience the viewer for yourself and be amazed by accessing it here! If you would like to access the code, please see the Github page for splaTV.

Featured

Recents

Featured

Research

Radiant Foam: Radfoam

Another novel Radiance Field representation is here with interesting capabilities.

Michael Rubloff

Feb 3, 2025

Research

Radiant Foam: Radfoam

Another novel Radiance Field representation is here with interesting capabilities.

Michael Rubloff

Feb 3, 2025

Research

Radiant Foam: Radfoam

Another novel Radiance Field representation is here with interesting capabilities.

Michael Rubloff

Platforms

Postshot adds Blackwell GPU Support

Postshot's most recent update has brought Blackwell support to the platform.

Michael Rubloff

Jan 29, 2025

Platforms

Postshot adds Blackwell GPU Support

Postshot's most recent update has brought Blackwell support to the platform.

Michael Rubloff

Jan 29, 2025

Platforms

Postshot adds Blackwell GPU Support

Postshot's most recent update has brought Blackwell support to the platform.

Michael Rubloff

Platforms

Brush 0.2 for 3DGS Released

The newest release from Brush is here, making it easier than ever to train 3DGS on Mac and beyond!

Michael Rubloff

Jan 29, 2025

Platforms

Brush 0.2 for 3DGS Released

The newest release from Brush is here, making it easier than ever to train 3DGS on Mac and beyond!

Michael Rubloff

Jan 29, 2025

Platforms

Brush 0.2 for 3DGS Released

The newest release from Brush is here, making it easier than ever to train 3DGS on Mac and beyond!

Michael Rubloff

Research

LinPrim: Linear Primitives for Differentiable Volumetric Rendering

A new Radiance Field representation has been published by the Technical University of Munich.

Michael Rubloff

Jan 29, 2025

Research

LinPrim: Linear Primitives for Differentiable Volumetric Rendering

A new Radiance Field representation has been published by the Technical University of Munich.

Michael Rubloff

Jan 29, 2025

Research

LinPrim: Linear Primitives for Differentiable Volumetric Rendering

A new Radiance Field representation has been published by the Technical University of Munich.

Michael Rubloff