Observations of Molnár's Interruptions
- Each line is centered on a lattice point of a grid
- Each line is rotated about its center
- Some lines are missing, mostly in clumps, not purely random
- The average rotation is vertical (or horizontal in one of them), and only some lines deviate very far from vertical
- The lines are arranged in a square with a side length of approximately 56
- The length of the lines is twice the distance between the centers of neighboring lines
- The amount that each line is rotated also appears to depend on location, with clumps of highly rotated lines and clumps of less rotated lines.
- The lines are black on a white background
- There is a blank margin approximately 2 to 3 lines long around the piece (and a date/number in the bottom left)
- The ratio of missing/present lines is small.
I started out with the observations listed above, and then coded the first draft. For the randomness of the rotations, after only a few modifications, I got to where it is now, which includes a random normal distribution multiplied with Perlin noise. However, getting the placement and quantity of missing lines took several more modifications to get where it is now. It has multiple layers of scaled, added and multiplied noise, including a pure random input to make it rougher around the edges. I was surprised to learn how many layers of randomness and noise were required to get the look I wanted. It's hard to tell after looking at so many iterations, but I think there is a more natural quality to Molnár's Interruptions compared to my own; I would be curious to see the algorithm she used.