← Overview · 3D gallery · GitHub

Asset library

11 assets composed into one USD stage, scan2usd_library.usda. Each asset is a payload (loaded on demand) with two variant sets authored in the library layer, so the asset files are never edited: collision (decomposed · convex_hull · bounding_box) and friction (measured · low · high). Total 8.21 kg, 345 convex colliders in the default variant.

Use it
scan2usd library output/assets -o library.usda        # build

from pxr import Usd
stage = Usd.Stage.Open("library.usda", Usd.Stage.LoadNone)   # instant: nothing loaded yet
mug = stage.GetPrimAtPath("/Library/ceramic_mug")
stage.Load(mug.GetPath())                                     # load just this asset
mug.GetVariantSet("collision").SetVariantSelection("convex_hull")
mug.GetVariantSet("friction").SetVariantSelection("low")

Why variants: the same scene can be simulated with cheap or accurate contact geometry (ablations, speed vs. fidelity), and contact properties can be randomized for robust policies, without copying assets.

ace_coffee_mug_kristen_16_oz_cup

Printed mug

size12.6 × 9.1 × 13.5 cm
mass466 g
inertia689 · 1032 · 1168 g·cm²
frictionμs 0.60 · μd 0.50
prim/Library/ace_coffee_mug_kristen_16_oz_cup
decomposed
convex decomposition · 32
convex_hull
one convex hull · 1
bounding_box
bounding box · 1
cole_hardware_bowl_scirocco_yellowblue

Ceramic bowl

size11.3 × 11.5 × 6.2 cm
mass192 g
inertia238 · 242 · 377 g·cm²
frictionμs 0.60 · μd 0.50
prim/Library/cole_hardware_bowl_scirocco_yellowblue
decomposed
convex decomposition · 32
convex_hull
one convex hull · 1
bounding_box
bounding box · 1
great_dinos_triceratops_toy

Android figure

size6.9 × 4.6 × 7.9 cm
mass104 g
inertia28 · 44 · 48 g·cm²
frictionμs 0.40 · μd 0.30
prim/Library/great_dinos_triceratops_toy
decomposed
convex decomposition · 11
convex_hull
one convex hull · 1
bounding_box
bounding box · 1
schleich_lion_action_figure

Lion figure

size11.5 × 4.2 × 7.0 cm
mass55 g
inertia15 · 61 · 67 g·cm²
frictionμs 0.40 · μd 0.30
prim/Library/schleich_lion_action_figure
decomposed
convex decomposition · 32
convex_hull
one convex hull · 1
bounding_box
bounding box · 1
womens_bluefish_2eye_boat_shoe_in_tan

Boat shoe

size24.9 × 9.0 × 9.0 cm
mass653 g
inertia515 · 2925 · 3018 g·cm²
frictionμs 1.00 · μd 0.80
prim/Library/womens_bluefish_2eye_boat_shoe_in_tan
decomposed
convex decomposition · 32
convex_hull
one convex hull · 1
bounding_box
bounding box · 1
ceramic_mug

Ceramic mug (synthetic scan)

size12.0 × 8.2 × 9.6 cm
mass372 g
inertia536 · 571 · 653 g·cm²
frictionμs 0.60 · μd 0.50
prim/Library/ceramic_mug
decomposed
convex decomposition · 25
convex_hull
one convex hull · 1
bounding_box
bounding box · 1
speckled_mug

Speckled mug (78 photos → USD)

size8.8 × 11.6 × 9.6 cm
mass436 g
inertia657 · 689 · 800 g·cm²
frictionμs 0.60 · μd 0.50
prim/Library/speckled_mug
decomposed
convex decomposition · 32
convex_hull
one convex hull · 1
bounding_box
bounding box · 1
keepsake_box

Hinged box (articulated)

size18.0 × 13.0 × 10.0 cm
mass539 g
inertiaarticulated (per body)
frictionμs 0.50 · μd 0.40
prim/Library/keepsake_box
decomposed
convex decomposition · 25
convex_hull
one convex hull · 2
bounding_box
bounding box · 2
buddha_head

Buddha head (67 real photos)

size15.2 × 16.2 × 20.0 cm
mass4624 g
inertia9024 · 13967 · 14135 g·cm²
frictionμs 0.70 · μd 0.60
prim/Library/buddha_head
decomposed
convex decomposition · 21
convex_hull
one convex hull · 1
bounding_box
bounding box · 1
cod_holder

Controller holder (own phone photos)

size10.1 × 10.8 × 21.0 cm
mass528 g
inertia443 · 1920 · 1970 g·cm²
frictionμs 0.40 · μd 0.30
prim/Library/cod_holder
decomposed
convex decomposition · 64
convex_hull
one convex hull · 1
bounding_box
bounding box · 1
controller

Game controller (own phone photos)

size14.6 × 13.5 × 5.6 cm
mass243 g
inertia151 · 410 · 511 g·cm²
frictionμs 0.40 · μd 0.30
prim/Library/controller
decomposed
convex decomposition · 39
convex_hull
one convex hull · 1
bounding_box
bounding box · 1

Renders: Scan2USD's numpy preview renderer. Third-party meshes: Google Scanned Objects (CC-BY 4.0); Buddha: AliceVision dataset (CC-BY 4.0).