Grid paintings à la Mondrian with JavaScript

March 04, 2017 | 9 minutes read

I was at a laundrette today and had just finished my book so I had some time to kill. Naturally I devised an algorithm for generating drawings that would resemble the grid-like paintings that Piet Mondrian made famous. With the benefit of hindsight I guess I could indulge in saner activities while waiting for my laundry to dry!

I went through different ideas but in the end I settled on a recursive approach. My idea is to divide a rectangle into two smaller ones and then to do the same with each sub-rectangle. Every time a rectangle is generated and is then filled with a random color; like Mondrian I use yellow, red, blue, black and white.

The work

In the end the algorithm I implemented generates images like these.

The code

The implementation is very short. I started by defining two classes - Point and Rectangle - for the sake of tidiness.

A rectangle is thus defined by an upper left point - called min - and a lower right point - called max. Here is an image just to make things clear.

The next step is to be able to draw a rectangle. The CanvasRenderingContext2D has a strokeRect(x1, y1, x2, y2) method but I had already written something before I knew this; anyway it isn’t very complicated. I attached the method to the Rectangle class.

I simply go around each corner of the rectangle in a clockwise fashion, starting at the top-left corner.

Kid’s stuff! Now comes the interesting part. The next step is to split a rectangle in two; there are two pitfalls to handle:

1. A rectangle can be half in two ways, either left/right or either top/bottom.
2. A split can produce rectangles which are too narrow and not visually appealing.

To handle the first pitfall, I split the rectangle into a left and a right rectangle if the rectangle’s width is greater than it’s height and vice-versa. To generate two sub-rectangles, I simply choose a random point along the $x$-axis in case of a left/right split or along $y$-axis in case of a top/bottom split.

To handle the second pitfall, I defined some padding numbers so that the points I generated were not too close to the current rectangle’s edges. The code for splitting a rectangle goes as follows:

It takes some time to get the point juggling right but all in all it’s pretty easy; the following illustration might help.

The randInt function simply generates an integer in the $[a, b)$ range, here is it’s definition:

As I mentioned the algorithm has to be recursive. This is easy to do, we only have to call the split method on each sub-rectangle. However the issue is to know when to stop. To fix this I simply introduced a depth parameter in the call to the split method; if the current depth is higher than a fixed limit then the recursion stops. Moreover, if a rectangle is too small - i.e. it not wide enough or not tall enough - then the recursion also stops. Finally I added the contouring and the filling of the rectangles into the split method. All in all the code is the following:

The colors variable has more whites so that the white color appears more often on the resulting image. To run the code with a canvas we simply have to obtain the canvas’s dimensions, create a rectangle from these dimensions and then call the split method on this initial rectangle.

That’s all there is to it! The full code is available on GitHub. I hope you enjoyed the read!