Subsets & Metadata

Pull only what you need — faster iterations, smaller storage.

Setup

pip install objaverse objaverse-plus

Filter by License & Category

import objaverse meta = objaverse.load_metadata() subset = [m for m in meta if m.get('license') == 'CC-BY' and 'chair' in m.get('categories', [])][:500] uids = [m['uid'] for m in subset]

Restrict by Polycount (pseudo-field)

subset = [m for m in meta if m.get('polycount', 0) <= 30000][:1000]

Neural‑only (Objaverse++)

import objaverse_plus as oplus neural = oplus.filter_neural(meta, types=['nerf','3dgs']) uids = [m['uid'] for m in neural][:200]
Acquire Objaverse.com