Why use 3D Gaussian Splatting when Matterport exists?

Quite often we hear this question, and it's a fair one. Let’s revisit it quickly. Matterport captures tripod-based, fixed-position panoramic images stitched into a dollhouse. Great overall quality, thanks to high-resolution photos, but corners and tight spots often stay hidden after stitching. That's an issue for inspection or detailed modelling.

3D Gaussian Splatting can build from various types of data: tripod shots similar to Matterport, free camera paths, drones, or robots. Afterwards, you can freely drop a virtual camera anywhere in the model and still clearly see the smallest details even in a corners.

A mesh from Matterport is possible, but without careful manual modelling, it won't fix visibility gaps. Before trusting any model, you probably want to confirm details with your own eyes. Gaussians leave that option open.

Curious how this looks on your data? Send it to us and we'll quickly run a demo, so you can judge yourself.