Gaussian Splat Objects Dataset

A small, hand-cleaned collection of everyday objects scanned as 3D Gaussian splats, free to download and use. Grown slowly, one object at a time

Marcel Padilla

ETH Zürich

Data on GitHub 📑 Cite

Most objects here are my own captures, scanned from a phone video with an open pipeline and cleaned up by hand. A few are made a different way: one generated by an AI model from a single image, or rendered from a known 3D asset and reconstructed. Each is clearly tagged so you always know which is which, and which license it carries. I know exactly where every one came from, so I can release them cleanly, with the details written next to it.

Use it

Install the loader and pull any object straight into numpy:

$ pip install padillasplats # in Python import padillasplats as ps s = ps.load("plant") # downloads once, caches, verifies, decodes s.positions # (113648, 3) float32 xyz s.colors # (113648, 4) uint8 RGBA for s in ps.load_all(category="object"): # loop the whole set print(s.id, len(s))

Filtering (ps.find(source_method="capture"), ps.find(tags=["plant"])) runs offline against a bundled inventory; only the splat bytes are downloaded. Full docs on GitHub.

Download

Pick which kinds you want. Each set downloads as one zip: the splats, their meta.json, and a LICENSES.txt that spells out the terms.

Loading the collection…

Or take the whole repository (data and metadata) as one zip, or use pip install padillasplats for scripted, filtered downloads.

The collection

Each object is tagged two ways: its Source (Scanned, a real capture; From geometry, converted or rendered from a known 3D mesh; or Generated, invented by an AI model from one image) and its Kind, object or scene. Filter by source with the chips below. Geometry objects ship in four levels of detail; pick the resolution to download from the dropdown in each row. The license sits under each Source: scanned and geometry objects are free to use (geometry ones are CC0 / public-domain classics), generated ones are provided as-is. Click a thumbnail to open it in the viewer.

Loading the collection…

	Object	Source	Kind	Size

Viewer

Drag to orbit, scroll to zoom, right-drag to pan.

Citing

If these are useful in your work, a citation is appreciated (and required by the data license):

Loading citation…

License

The license depends on how an object was made (it is shown under each Source in the table, and per object in its meta.json):

Scanned objects are CC-BY-4.0. Use them for anything, including commercially. You must credit the author.
Rendered objects take the license of the 3D model they were rendered from (the rendered bunny is CC-BY-3.0, from the Stanford Bunny STL by Makerbot via Wikimedia Commons). Credit that original author as well.
Generated objects are provided as-is, with no warranty of rights. They were invented by an AI image-to-3D model whose training data is undisclosed, so their copyright status cannot be warranted and they are not under CC-BY. You are responsible for clearing any rights for your use.
The loader code is 0BSD. No attribution required.

Credit line you can paste for a CC-BY object: Gaussian splat by Marcel Padilla, from github.com/marcelpadilla/splats, licensed CC-BY-4.0. The license covers the capture, render or generation and the reconstruction, not the design of any object depicted.