In the history of art, I am primarily interested in color, as the most important element in the development of this history. The color palette of painting is constantly expanding in time, space, styles and temperaments, imagination and materials used. If we concentrate on the number of paintings created throughout history (I found a figure of 15 billion), then how to realize, to assimilate this grandiose scale?
At first there appeared a simple curiosity. Is it possible to imagine all this mass of canvases as a single color composition? We have to apply a common denominator to each picture in the form of an average color, we will see a mosaic consisting of multi-colored pixels. But the task of processing billions of paintings seems overwhelming without some kind of sampling.
The principle must be completely randomized. For the beginning, let's take artists whose last name starts with the letter A and select one painting of each artist (the first one in a Google search). Belonging to time, genre, art school and other color determining features will be completely excluded. The result is a 3600, or 60x60 pixel composition representing artists starting with the letter A.
Now let's repeat the same with the letter Z. Here we have 900, or 30x30 pixels. But we see that the overall coloring remains unchanged. You can extrapolate the result and conclude that all other letters will give a similar flavor with a difference only in the number of pixels. Topic closed. Issue resolved. Is not it?
Let's complicate the issue. How to highlight a large amount of work with any features that affect the result? The history itself suggests the principle of dividing artists by style. Let's compare Byzantine art, Renaissance, Baroque, Impressionism, Cubism, Fauvism, Conceptual art and other movements. The difference is clear and multifaceted. You can draw conclusions and analyze.
But let's dig even deeper. Museums. Do long-term collections, biases of different selection committees and curators affect the color of pixel compositions? For even greater “beauty” of the experiment, I choose 3 museums in New York (Metropolitan, Guggenheim, MOMA) and 3 museums in Paris (Louvre, d’Orsay, Center J. Pompidou). There is a difference. Not so striking. But it gives food for thought.
Let's take a breath. Luckily, the project took several years to complete. How do political leanings affect art? Politicians classify artists as one group or another. We remember the "degenerate art" during the third Reich and can compare it with the official creations (small but manageable difficulty in finding work) of the same period. A couple of Soviet and nonconformist arts in the USSR is also interesting. Influence and very eloquently.
And the last (for today) experiment. It is not about finding differences, but about identifying commonality of the color pixel, despite ... the differences of nations. In this case, it is necessary to comply a single algorithm for selecting pictures and their color averaging. The first is a selection of 100 countries with a sufficient number of artists in Google search. Second - each country is represented by 35-50 authors, one (the first in the search display) picture from each artist. Third - step by step averaging of each work, transforming into a same size pixel and averaging all 35-50 pixels into one, then placing them in alphabetical order. 100 (10x10) pixel-nations with fairly similar hues.
Does this mean that there are national differences in general color? No, a more complete sample is needed for such verification. Perhaps we need all artists of every country and all of their works. Also this result will depend on the changes in color reproduction when working in Photoshop by millions of photographers and a displacement of this mass of photos across the expanses of Google.
