Dennis0626/3d-1k-assets

3D-1K Assets

A curated 3D rigid-object asset library with a 10-category taxonomy and 1,000 GLB assets (one per base object class). Generated by a text → image → RGBA → 3D pipeline (FLUX.2-klein → RMBG-2.0 → TRELLIS.2).

This is the base class layer of a planned 1,000 × 1,000 = 1,000,000 instance dataset for embodied AI / robotic manipulation.

Files

file	size	content
glbs.tar	6.5 GB	1000 GLB assets: 0001.glb … 1000.glb
rgba_images.tar	700 MB	1000 RGBA seed images (foreground extracted)
seed_images.tar	615 MB	1000 raw RGB seed images (pre-background removal)
renders.tar	102 MB	Multi-view renders <id>_{front,back,left,right}.png (sparse 32-id sample)
metadata.json	—	Taxonomy + per-asset description & prompt
scales.json	—	Per-asset physical scale (meters)
classes.py	—	Source taxonomy definition

After download, extract:

mkdir glbs rgba_images seed_images renders
tar -xf glbs.tar -C glbs/
tar -xf rgba_images.tar -C rgba_images/
tar -xf seed_images.tar -C seed_images/
tar -xf renders.tar -C renders/

Taxonomy (10 categories, 1000 items)

category_id	category_en	category_zh	count
1	Kitchenware & Dining	厨房与餐饮具	150
2	Food & Packaging	食品与包装	150
3	Household & Living	家居与生活用品	120
4	Office & Electronics	办公与电子产品	100
5	Hardware & Tools	工具与五金	100
6	Toys & Hobby	玩具与娱乐	80
7	Medical & Care	医药与个人护理	60
8	Industrial & Lab	工业与实验室	80
9	Apparel Accessories	衣物配件（硬质）	60
10	Miscellaneous & Abstract	杂项与抽象几何体	100

metadata.json schema

{
  "id": "0001",
  "category_id": 1,
  "category_en": "Kitchenware & Dining",
  "category_zh": "厨房与餐饮具",
  "name_en": "ceramic soup bowl",
  "name_zh": "陶瓷汤碗",
  "material": "ceramic",
  "key_features": "wide rim, round profile, shallow depth",
  "prompt": "A ceramic soup bowl, ..."
}

id (zero-padded 4-digit) is the join key across all archives and scales.json.

Usage

from huggingface_hub import snapshot_download
path = snapshot_download("Dennis0626/3d-1k-assets", repo_type="dataset")

License

Released under Apache 2.0. The GLB assets are AI-generated; verify suitability for your downstream use case.