Visualise a packing

Use this page when you want to look at a result rather than read its numbers. packingsolver3d.visual draws a Result the way upstream’s own scripts/visualize_box.py does – a translucent bin, one opaque cuboid per item with its type id, one 3D scene per bin – and returns a plotly figure you can rotate in a browser or notebook.

Install the optional dependency

pip install "packingsolver3d[plot]"      # or: pip install plotly

Nothing else in the package needs plotly; importing packingsolver3d.visual without it raises ImportError with this hint.

Draw a result

from packingsolver3d import BinType, Instance, ItemType, Objective, box
from packingsolver3d.visual import plot_result

instance = Instance(
    bin_types=[BinType(x=100, y=100, z=100, cost=10, copies=5)],
    item_types=[ItemType(x=20, y=30, z=40, copies=6), ItemType(x=15, y=15, z=15, copies=4)],
    objective=Objective.BIN_PACKING,
)
result = box.solve(instance, time_limit=2.0)

figure = plot_result(result)          # one colour per item type, ids written on the boxes
figure.show()                          # opens a browser tab, or renders inline in a notebook
figure.write_html('packing.html')      # self-contained interactive file

Colour by stack

For boxstacks results, color_by='stack' gives every stack its own colour so the piles stand out; color_by='same' draws everything in one colour and show_ids=False drops the labels when there are hundreds of items.

from packingsolver3d import boxstacks
from packingsolver3d.visual import plot_result

figure = plot_result(boxstacks.solve(instance, time_limit=2.0), color_by='stack')

Several bins

Every bin gets its own scene; the scenes are laid out in a grid of ceil(sqrt(bins)) columns, or as many as columns asks for. plot_bin() draws a single PackedBin.

Export to PNG

Static images go through plotly’s write_image, which needs kaleido (version 1 or newer) and a Chrome or Chromium binary on the machine:

pip install "kaleido>=1"
figure.write_image('packing.png', width=900, height=650, scale=2)

The figures on these pages were produced exactly this way by tools/make_figures.py in the repository.