• The background is a white square
  • There is a white margin around the outside
  • The piece is composed of many black lines, each the same length
  • There are 52 lines in one row or column of the square
  • The positioning of the midpoints of the lines is regular, their midpoints are all placed equal distances apart
  • The lines aren’t entirely randomly rotated, they’re sort of “straight-ish” with a lot of random variation
  • Any two adjacent lines are about one half of their height apart from each other
  • There are blank areas without any lines in them
  • The blank areas vary in size throughout the piece, but there seems to be some form to them
  • Within the blank areas there are still scattered lines here and there
  • Some of the generated images have more or larger blank areas than others
  • The smallest blank areas have only one or two lines missing
  • The largest blank areas I would estimate have about 12 lines missing
  • In any given piece, there are no more than a few areas of interruption
  • The “interrupted” areas seem roughly rectilinear, but they don’t look it because of the messy lines around them
  • Despite the larger areas of interruption, there are smaller ones too that are harder to notice

My reproduction – GIF

My Reproduction – Interactive


I started by generating a grid of randomly rotated lines, but quickly realized that the rotations of the lines weren’t completely random, they were usually close to vertical. To replicate her work more accurately, I made the lines tend towards being straight up and down according to a normal distribution. For the interruptions, each time the work is regenerated, I decide how many holes to create and then recursively flood-fill outwards from the starting point of the hole with 50% probability of continuing and 50% probability of stopping.

I tweaked it as much as I could to replicate a Molnar-like effect, and tried implementing recursion and randomness wherever I could, but the process still felt over-engineered compared to the beauty of what she did. It makes me wonder what her code looked like. In doing this assignment, I gained a lot of appreciation for the thought behind each piece of Molnar’s work. The parts of the piece that seem easy to take for granted, such as the fact that the holes are semi-contiguous shapes, each require specific logic to execute.