Message Boards Message Boards

More Shape-Shifting in 3D based on Intersecting Cylinders

Posted 6 years ago

MODERATOR NOTE: a submission to computations art contest, see more: https://wolfr.am/CompArt-22


As was illustrated in the latest Wolfram Insider, "shape-shifting" and ambiguous views of space curves can be derived from the intersection of perpendicular cylinders.

My recent Wolfram Community contribution More on Intersecting Cylinders and "Ambiguous Rings" [1], gives the conditions to obtain fully closed intersections curves between the cylinders. A "tight fit" between the cross section of the polygonal cylinder and the longitudinal cross section (two parallel lines) of the circular cylinder.

This requires the radius of the circular cylinder and the offset between the cylinders adapted to the axial rotation of the polygonal cylinder. For regular polygons, this tight fit can easily be computed with some elementary analytical geometry. These are the tight fit conditions for a triangular cross section and an animation to verify:

fittedRadius3[t_] := 
 1/2 (Max[-Cos[1/6 (\[Pi] - 6 t)], Cos[\[Pi]/6 + t], Sin[t]] - 
    Min[-Cos[1/6 (\[Pi] - 6 t)], Cos[\[Pi]/6 + t], Sin[t]])
fittedOffset3[t_] := 
 1/2 (Max[-Cos[1/6 (\[Pi] - 6 t)], Cos[\[Pi]/6 + t], Sin[t]] + 
    Min[-Cos[1/6 (\[Pi] - 6 t)], Cos[\[Pi]/6 + t], Sin[t]])
Animate[ParametricPlot[
  1/2 Sec[2/
     3 ArcTan[Cot[(3 (\[Theta] - \[Theta]0))/2]]] {Cos[\[Theta]], 
    Sin[\[Theta]]}, {\[Theta], 0, 2 \[Pi]}, 
  PlotStyle -> Directive[AbsoluteThickness[2], Blue], 
  PlotRange -> 1.25, Background -> Black, ImageSize -> Small, 
  TicksStyle -> 6, 
  Prolog -> {{AbsoluteThickness[1], Dotted, Circle[]}, Red, 
    AbsoluteThickness[2], 
    Line[{{{-5, 
        fittedRadius3[\[Theta]0] + fittedOffset3[\[Theta]0]}, {5, 
        fittedRadius3[\[Theta]0] + 
         fittedOffset3[\[Theta]0]}}, {{-5, -fittedRadius3[\[Theta]0 + \
\[Pi]] - fittedOffset3[\[Theta]0 + \[Pi]]}, {5, -fittedRadius3[\
\[Theta]0 + \[Pi]] - fittedOffset3[\[Theta]0 + \[Pi]]}}}], Red, 
    AbsolutePointSize[5], Point[{0, fittedOffset3[\[Theta]0]}], Black,
     Point[{0, 0}]}], {\[Theta]0, 0, 2 \[Pi]}, 
 TrackedSymbols :> Manipulate]

tightfitrose5

Entering this tight fit radius and -offset into the polyRingsetCF function from [1], we can make an animation scrolling through all possible closed intersections of a triangular and a circular cylinder:

Animate[Module[{n = 3, \[Alpha] = 0. r, d},
  r = fittedRadius3[\[Theta]0];
  d = fittedOffset3[\[Theta]0];
  ParametricPlot3D[
   polyRingsetCF[t, r, \[Theta]0, d, 
    n, \[Alpha]], {t, -\[Pi], \[Pi]},
   PlotStyle -> {{Green, Tube[.035]}}, PerformanceGoal -> "Quality", 
   SphericalRegion -> True, PlotRange -> 6, 
   Background -> Lighter[Gray, 0.5], ViewPoint -> .7 {3, -3, 3}, 
   ViewAngle -> 4 \[Degree], Boxed -> False, Axes -> False]],
 {\[Theta]0, 0, 2 \[Pi]}]

closedroserings3

One can extend the idea of closed intersection curves showing shape-shifting to other than polygonal cylinders. An interesting case are intersections of rose-curve based cylinders (see my Wolfram Demonstration: "Ambiguous Rings Based on a Rose Curve" [2]. Here again, to a have tight fit of the cylinders, requires a radius and axial offset of the circular cylinder adapted to the axial rotation of the rose based cylinder.

Only in this case, it is not that simple to find an analytical solution. But we can find a numerical solution using the function FindMaxvalue to find the tangents for the parallel lines to the rose curve and build an interpolating function.

roseOf5[\[Theta]_, \[Theta]0_: 0](*parametric of a 5-petaled rose \
curve*):= {1/3 Cos[\[Theta]] (2 + Sin[\[Theta]0 + 5 \[Theta]]), 
  1/3 Sin[\[Theta]] (2 + Sin[\[Theta]0 + 5 \[Theta]])}
tightFits = 
  Table[{\[Theta]0, 
    FindMaxValue[Last@roseOf5[\[Theta], \[Theta]0], \[Theta], 
     PrecisionGoal -> 12, 
     WorkingPrecision -> 16]}, {\[Theta]0, -\[Pi], \[Pi], \[Pi]/24}];
tightFitIF[\[Theta]0_] := Module[{f}, f = Interpolation[tightFits];
  Piecewise[{{f[Mod[\[Theta]0, 2 \[Pi]]], 
     0 <= Mod[\[Theta]0, 2 \[Pi]] < \[Pi]}}, 
   f[Mod[\[Theta]0, 2 \[Pi]] - 2 \[Pi]]]]
fittedRadiusRose5[t_] := (tightFitIF[t] + tightFitIF[t + \[Pi]])/2;
fittedOffsetRose5[t_] := (tightFitIF[t] - tightFitIF[t + \[Pi]])/2;

The tight fit can be demonstrated with an animation:

Animate[Quiet@
  ParametricPlot[roseOf5[\[Theta], \[Theta]0], {\[Theta], 0, 2 \[Pi]},
    PlotStyle -> Directive[AbsoluteThickness[4], Blue], 
   PlotRange -> 1.25, 
   Prolog -> {{Thick, Dotted, Circle[]}, Red, AbsoluteThickness[4], 
     Line[{{{-5, 
         fittedRadiusRose5[\[Theta]0] + 
          fittedOffsetRose5[\[Theta]0]}, {5, 
         fittedRadiusRose5[\[Theta]0] + 
          fittedOffsetRose5[\[Theta]0]}}, {{-5, -fittedRadiusRose5[\
\[Theta]0] + 
          fittedOffsetRose5[\[Theta]0]}, {5, -fittedRadiusRose5[\
\[Theta]0] + fittedOffsetRose5[\[Theta]0]}}}], AbsolutePointSize[5], 
     Point[{0, fittedOffsetRose5[\[Theta]0]}], Black, 
     Point[{0, 0}]}],
 {\[Theta]0, 0, 2 \[Pi]}]

tightfitrose5e

Using the roseRingsetCF function from [2], we get an animation scrolling through all possible closed intersections of a 5 petaled rose curve based cylinder and a circular cylinder :

Animate[Quiet@Module[{n = 5, \[Alpha] = 0, d, r},
   r = fittedRadiusRose5[\[Theta]0];
   d = fittedOffsetRose5[\[Theta]0];
   ParametricPlot3D[
    roseRingsetCF[t, r, \[Theta]0, d, n, \[Alpha]], {t, -\[Pi], \[Pi]},
    PlotStyle -> {{Lighter[Red, .2], Tube[.04]}}, 
    SphericalRegion -> True,
    PlotRange -> 6, Background -> Lighter[Gray, 0.65], 
    ViewAngle -> 4 \[Degree], Boxed -> False, Axes -> False]],
 {\[Theta]0, 0, 2 \[Pi]}]

enter image description here

We can now put both animations together and demonstrate yet another example of shape-shifting in 3D:

enter image description here

This shows the "ambiguous" views that can be obtained while rotating the viewpoint around one of these 'shifted" shapes. Notice the circle and the 5-petaled rose views when the rotation is stopped briefly. These two views are at perpendicular view directions.

enter image description here

Selecting only half of this ringset will give an "ambiguous ring". This is a ring that can be seen as a rose or a circle, depending on the view direction. Two, almost perpendicular, view directions can be achieved by means of a mirror. An object in front of a mirror can be viewed either directly or reflected in the mirror. The two view directions form an almost right angle.

enter image description here

This is the code to generate and 3D-print these two ambiguous rings:

Module[{d, r, t1, 
   t2, \[Theta]0}, {d, r, t1, t2, \[Theta]0} = {0, 
    0.956394658621, -0.58857, 0.58857, -\[Pi]/2};
  roseRing51 = 
   ParametricPlot3D[
    roseRing1CF[t, r, \[Theta]0, d, 5, 0., t1, t2], {t, -\[Pi], \[Pi]},
    PlotStyle -> {{Lighter[Red, .2], Tube[.065]}}, PlotPoints -> 200, 
    PerformanceGoal -> "Quality", SphericalRegion -> True,
    PlotRange -> 1.08, Background -> Lighter[Gray, 0.65], 
    ImageSize -> 300, Boxed -> False, Axes -> False]];
Printout3D[roseRing51, "Sculpteo", TargetUnits -> "Centimeters", 
 RegionSize -> 5]

enter image description here

Module[{d, r, t1, 
   t2, \[Theta]0}, {d, r, t1, t2, \[Theta]0} = {0, 
    0.956394658621, -0.58857, 0.58857, -\[Pi]/2};
  roseRing51 = 
   ParametricPlot3D[
    roseRing1CF[t, r, \[Theta]0, d, 5, 0., t1, t2], {t, -\[Pi], \[Pi]},
    PlotStyle -> {{Lighter[Red, .2], Tube[.065]}}, PlotPoints -> 200, 
    PerformanceGoal -> "Quality", SphericalRegion -> True,
    PlotRange -> 1.08, Background -> Lighter[Gray, 0.65], 
    ImageSize -> 300, Boxed -> False, Axes -> False]];
Printout3D[roseRing51, "Sculpteo", TargetUnits -> "Centimeters", 
 RegionSize -> 5]

enter image description here

Now it is time to try these two ambiguous rings in front of a mirror and see the direct and the reflected views simultaneously. For each ring, we have two options :

  1. direct view is a 5-petaled rose and reflected view is a circle (right photo)

  2. direct view is a circle and reflected view is a 5-petaled rose (left photo)

enter image description here

enter image description here

The number of such "ambiguous rings" is almost unlimited and depends only on the cross sections of the intersecting cylinders. maybe the readers of this community can try their own unique combination!

POSTED BY: Erik Mahieu
4 Replies
Posted 6 years ago

The two additional arguments t1 and t2 are the angles t in the ParametricPlot3D that will split the ring"set" into 2 separate "rings". By lack of finding a formula, I determined these two angles experimentally by means of another Mathematica notebook With a Manipulate introducing the two new variables t1 and t2 this was very easy. See the notebook attached in which the complete code for the rings is included

Attachments:
POSTED BY: Erik Mahieu

roseRingsetCF might be defined something like?

roseRingsetCF=Compile[{{\[Theta], _Real},{r, _Real},{\[Theta]0, _Real},{d, _Real},{n, _Integer},{\[Alpha], _Real}},Module[{t1,t2},(*2 part composite curve*)t1=Sec[2 ArcTan[Cot[n (\[Theta]-\[Theta]0)/2]]/n];
t2=Sec[\[Alpha]] Sqrt[-d^2+r^2+2 d Cos[\[Pi]/n] t1 Sin[\[Theta]]-Cos[\[Pi]/n]^2 t1^2 Sin[\[Theta]]^2];
{(*part1*)roseRing1CF[\[Theta],r,\[Theta]0,d,n,\[Alpha],t1,t2],(*part 2*)roseRing2CF[\[Theta],r,\[Theta]0,d,n,\[Alpha],t1,t2]}]];

Michael

POSTED BY: Michael Kelly

There is no function roseRingsetCF function defined in [2]. Instead there are functions roseRing1CF and roseRing2CF. However these functions take 8 arguments whereas roseRingsetCF takes only 6 arguments. I understand that roseRingsetCF will probably be defined along the lines of

Through[{roseRing1CF,roseRing2CF}[##]]&[t,r,\[Theta]0,d,n,\[Pi]/2+\[Alpha],t1,t2]

but this still leaves unspecified the last two arguments t1 and t2. Are they to be defined like t and s in the definition of polyRingsetCF ?

Regards

Michael Kelly

POSTED BY: Michael Kelly

enter image description here - Congratulations! This post is now a Staff Pick as distinguished by a badge on your profile! Thank you, keep it coming!

POSTED BY: EDITORIAL BOARD
Reply to this discussion
Community posts can be styled and formatted using the Markdown syntax.
Reply Preview
Attachments
Remove
or Discard

Group Abstract Group Abstract