In the past, I’ve experimented with various programatic methods that dealt with image processing —- eventually I learned of new areas to critically engage with digital artifacts and reconstruct them in interesting and surprising ways. This led to an idea.

What if all the imagery from an artists body of work was condensed in some fashion? Could the information be aggregated into one image?

The base imagery for this body of work came from a dataset called the Best Artworks of All Time (please ignore the completely subjective title— it is actually a great collection of artists and images of their digital collections).

Methodology

One way to achieve this goal is to focus on the RGB values within the pixels themselves. Each pixel has formatted information—a list of three values within brackets called a tuple.

Which looks like:

[255, 252, 230]

The first python script takes the RGB value of each pixel for the artists collection of images (around 150-200 for each artist) and calculates the average at that specific position. Below is an example.

PIXEL 1 IMG 1 [255, 220, 230] + PIXEL 1 IMG 2 [255, 221, 230] + PIXEL 1 IMG 3[255, 234, 230] = [765, 675, 690]

[765, 675, 690] / 3 = FINAL PIXEL [255, 225, 230]

The second script uses image segmentation and contour functions to trace areas of the artist’s works. It does this for each image and combines them all into one plot.

The first image is a culmination of all the final pixel averages of each artists collection.

The second image is akin to if you traced all the elements of each painting onto one piece of paper.