Beer pong

Basic contacts

Throw a ping-pong ball: one bounce, then into the cup.

Play0.00 s

¼×½×1×2×

Loop

The steel preset (the default), recordedon an NVIDIA RTX PRO 6000 Blackwell.

Run it

cd examples
# the default preset (steel), then the others
uv run shelf play beer_pong
# change solver settings live
uv run shelf tune beer_pong

Source: examples/shelf/01_basic_contacts/beer_pong/ · ported from steel-webgpu assets/scenes/beer_pong (hero_montage export) @ ad03e49

Presets

Preset	Solver	Status	Checked
steeldefault	steel	works	2026-11-10, newton f6fc82a8b, NVIDIA RTX PRO 6000 Blackwell Server Edition

Solvers with no row are untested on this example. The full settings are inpresets.toml.

Code

example.py

examples/shelf/01_basic_contacts/beer_pong/example.py

# SPDX-FileCopyrightText: Copyright (c) 2026 Skild AI and the Steel contributors
# SPDX-License-Identifier: Apache-2.0
# Direct source: steel-webgpu; see presets.toml and THIRD_PARTY.md.
# Modified for Steel Shelf: preset-driven execution and shared runner integration.

###########################################################################
# Example Beer pong
#
# The arm picks up a ping-pong ball and throws it; it bounces once on the table and lands in the front cup.
#
# Rebuilt from steel-webgpu's exported scene (assets/scenes/beer_pong: scene.json
# and scene.bin, copied next to this file); see common/exported_scene.py. The
# visual meshes draw when a steel-webgpu checkout sits next to the shelf.
#
# Command: python -m shelf run beer_pong
#
###########################################################################

import newton.examples
from common.exported_scene import ExportedSceneExample


class Example(ExportedSceneExample):
    file = __file__


if __name__ == "__main__":
    parser = Example.create_parser()
    viewer, args = newton.examples.init(parser)
    newton.examples.run(Example(viewer, args), args)

example.py on GitHub

presets.toml

examples/shelf/01_basic_contacts/beer_pong/presets.toml

example = "beer_pong"
source = "steel-webgpu assets/scenes/beer_pong (hero_montage export) @ ad03e49"
default = "steel"

[check]
# Arguments `python -m shelf check` passes in test mode.
args = []

[gallery]
short = "Throw a ping-pong ball: one bounce, then into the cup."
title = "Beer pong"
frame = 300

[xpbd]
status = "untested"

[vbd]
status = "untested"

[mujoco]
status = "untested"

[featherstone]
status = "untested"

[semi_implicit]
status = "untested"

[feather_pgs]
status = "untested"

[steel]
status = "works"
# 10 substeps x 12 iterations, immediate response: the cheapest setting that holds the 2.7 g ball without the
# fingers sinking into it. The fingers (0.08 kg armature) squeeze the ball 5 mm past its surface; Gauss-Seidel
# converges slowly at that mass ratio, so at the browser's 4 x 16 the pads sit 4.9 mm inside the ball (visible)
# and the throw leaves 28 mm off a converged 16 x 32 run. Here 1.7 mm and within 5 mm of it, and the ball bounces
# once and lands in the front cup. 8 x 12: 2.5 mm; 16 x 4 / 16 x 6 miss the cup. Immediate response is ~2x
# faster than propagation_colored at equal behavior. Velocity limits off: every exported limit is 1e6.
# Rows: at most 36 used; 128 leaves headroom. Re-checked with the finger mimic (cfeff10): unchanged (the grip was
# already centred); 8 x 16 then grips 1.8 mm but its throw leaves 17 mm off the converged run on Blackwell.
sim = { fps = 60, substeps = 10 }
solver = { solve_mode = "matrix_free", iterations = 12, articulated_contact_response = "immediate", enable_joint_limits = true, joint_limit_activation_gap = 0.1, enable_joint_velocity_limits = false, articulation_constraint_max = 128, free_body_constraint_max = 128 }
checked = "2026-11-10, newton f6fc82a8b, NVIDIA RTX PRO 6000 Blackwell Server Edition"

[kamino]
status = "untested"

[implicit_mpm]
status = "untested"

[style3d]
status = "untested"

presets.toml on GitHub

← Basic conveyor forcesAll examplesBounce →