Dropped Call

The Dirichlet distributions are a family of distributions on the $n$-simplex $x_1 + \dots + x_{n+1} = 1$. For example, the Dirichlet$(1,\dots , 1)$ distribution gives the uniform distribution on the simplex. Of course, generalized (normalized) barycentric coordinates on any convex polytope also sum to 1, so one can get an induced distribution on a convex polytope (of course, one way to think of generalized barycentric coordinates is as defining a map from a simplex to the polytope, so equivalently we're talking about the pushforward distribution on the polytope under this map).

In any case, each frame of the animation is produced by generating generalized barycentric coordinates on the square according to the Dirichlet$(5/4,5/4,5/4,5/4)$ distribution and then creating a DensityHistogram[] of the points. Here's the code for producing a single frame:

Module[{n = 200000, verts, dirichletvars},
 verts = {{0, 0}, {0, 1}, {1, 1}, {1, 0}};
 dirichletvars = 
  Append[#, 1 - Total[#]] & /@ 
   RandomVariate[DirichletDistribution[Table[5/4, {Length[verts]}]], n];
 DensityHistogram[Table[dirichletvars[[i]].verts, {i, 1, n}], {.025}, 
  Frame -> False, ColorFunction -> GrayLevel, 
  PlotRangePadding -> None, Background -> Black, ImageSize -> 540]
 ]