1 / 6
Crooked Dice — a fair die, a gaff die and a loaded die that look identical — image 1
1 / 6

Crooked Dice — a fair die, a gaff die and a loaded die that look identical

Order a print of this model

We calculate the final price from your selected file. You review and accept it before payment.

Saved on this device.

Three 16 mm d6 from one source: an honest die, a duplicate-face gaff showing only 5 and 6

Rating
0 / 5
Likes
0
Downloads
6
Makes
0
Views
51
Reviews
0

Updated at source:

Model information

Category
Board Games
License
Creative Commons — Attribution

Three 16 mm d6 that are identical from the outside — same body, same size, same pip style, same weight to within a tenth of a gram — and a cup that tells them apart.

die

what it is

fair

honest — 1…6, opposite faces sum to 7, right-handed (1 up, 2 toward you, 3 on the right)

gaff

duplicate-face die, showing only 5 and 6. Every face uses a standard pip pattern, so a glance reads "ordinary die"

loaded

honest pips, dishonest inside — a sealed void under the 6 face, with an optional pocket for a steel slug

Plus the float-test cup: the counter-measure that catches the loaded die. That is the honest framing and the interesting one — a party and close-up-magic prop, sold with the way to defeat it in the same download.

It tells you how crooked it is, as a number

Loaded dice are sold with an adjective ("heavily weighted!") and never with a figure. These come with two, both measured off the meshes you are downloading:

  • CoM offset — how far the centre of mass sits from the geometric centre.

  • Tipping barrier per face — a die resting on a face pivots about one of its edges, and its centre of mass has to climb to get over that edge. Six equal barriers is a fair die; the max/min ratio is the honest way to state how strong a load is.

die

CoM offset

barrier ratio

most stable rest → shows

fair, engraved

0.0426 mm (0.53 %)

1.008

—

fair, pips filled flush

0.0000 mm

1.000

—

gaff 5/6, engraved

0.0126 mm (0.16 %)

1.001

—

loaded, sealed void

0.2614 mm (3.27 %)

1.080

rests on 1 → shows 6

loaded + empty slug pocket

0.0638 mm

1.015

(nearly fair)

loaded + steel slug seated

1.0273 mm (12.8 %)

1.391

rests on 1 → shows 6

Two things fall out of that table that are worth saying out loud.

An ordinary engraved die is already slightly crooked. Six pips remove more material than one, so the 1 face is the heavy side and the 6 is the light one — 0.53 % of the half-width here. Filling the pips flush with the same polymer puts it back: the flush-filled die measures 0.0000 mm off centre. That is why casino dice are paint-filled, and it means the two-colour print is the fairer of the two.

The slug version leaves the printer nearly fair. The pocket removes mass from the side opposite the void and very nearly cancels it — 0.0638 mm, barrier ratio 1.015. The die only becomes strongly loaded once the steel goes in. Which is also why "weigh it" does not catch it, and the float test does.

What the barrier ratio is not: a probability. How often a face lands up depends on the throw, the bounce and the surface. The ratio says which faces are more stable and by how much — the true statement that geometry alone supports.

Two colours

crooked_die_fair_2c.stl and crooked_die_gaff_2c.stl are the die split into two solids: the body, and exactly the material the pips take out of it. In Bambu Studio: Import → right-click → "Split to Parts", then give pips the second filament. Two parts, coordinates exact. Print the _2c file instead of the plain one, not as well as.

Twenty-one pips would be twenty-one separate parts to colour one at a time, so the ink is welded to itself under the skin first — a pad beneath each face's cluster and three rods through the centre of the die, on the face normals. The rods are symmetric and every pad is the same thickness, because this die's whole claim is that its balance is measured: the web leaves the ink less lopsided than the bare pips (1.786 mm against 2.155 mm), it stops at the pip floor so none of it reaches a surface, and body + ink still sums to the blank exactly.

The loaded die has no _2c file, on purpose. It carries a sealed void, and Split to Parts hands a void back as a positive part — which would silently fill the load and print you a fair die. Print crooked_die_loaded.stl in one colour, or paint-fill the pips.

Files

file

what

crooked_die_fair.stl

honest die, one colour

crooked_die_gaff.stl

duplicate-face 5/6 die, one colour

crooked_die_loaded.stl

sealed void under the 6

crooked_die_loaded_slug.stl

same, with a Ø8 mm slug pocket

float_cup.stl

the float-test cup that catches it

crooked_die_fair_2c.stl

two-colour fair die — body + pips, Split to Parts

crooked_die_gaff_2c.stl

two-colour gaff die

crooked_dice.scad

the full parametric source

Printing

Any face on the bed. No supports, no brim, ≥ 40 % infill — uneven infill is what makes a printed die roll badly, and on the loaded die it is also what the bias is made of.

The sealed void is a bridge over open air. Do not run a mesh "repair" that removes internal shells, and do not let a Remove internal cavities type option touch it — that silently prints a fair die.

Slug version: the pocket takes a Ø8 mm steel ball or an M8 nut. Pause the print at the layer where the pocket roof begins (Bambu Studio → right-click the layer → Add pause), drop the slug in, resume. Read that height off the sliced preview, not off the model — it is a print height, not a model height.

What this is for

A magic and party prop, and a demonstration of how a loaded die is caught. It is not advice for playing with other people's money.

Source

Hand-authored parametric OpenSCAD, self-contained, attached. size_mm, corner_r, pip_d, pip_depth, pip_pitch, plus mode="fair"|"gaff"|"loaded", gaff_values, load_face, void_d / void_h / void_offset, slug=true / slug_d and the pip_web parameters for the two-colour weld. The corner-sphere facet count is treated as a dimension rather than a smoothness setting — the flats are tangent planes, so a coarse value lands the width short of 16 mm.

Tags