Working with data and creating generative art can be beautiful, unpredictable and meaningful and also not obvious and lacking substance. What a funny balance. When you stack a bunch of images on top of each other (shorthand explanation of image averaging) they all kind of blob together. Sometimes there’s meaning in the chaos and sometimes it’s just chaos.

My favorite from this project is this one, above, which is all my photographs taken on vacation traveling and hiking through Maine in the summer of 2021. For me, this mixture of colors, greens and light blues, and the shapes which feel like a forest or a mountain, add up to a abstract Maine vacation. Of course, it’s also a happy memory which plays a factor in why I particularly like it.

But simply combining a bunch of images doesn’t always add up to something interesting. Take this image which is every picture I took while some friends visited me in Philadelphia. This gobbledygook does not represent that delightful weekend at all. It really doesn’t represent anything. You can’t even tell we ate pizza on Friday!

We have to be a little careful with data and art. Projecting meaning when all we really have is a pixel smoothie is an easy trap. 

But other times I think something compelling can be extrapolated. Here above are all the US president portraits up until Donald Trump combined. There is a figure in this. Perhaps presidents in portraits stand the same way and stare the same way? Look the same way? Are the same way? I also recognize quite clearly that the figure in this image is clearly lighter in complexion. All, except one of the US presidents have been white. This experiment, in a way heightens or highlights something important but difficult to articulate about American society, about leadership and who we choose to govern. When doing data and algorithmic art I believe the artist should have an hypothesis prior to their experiment. What might the sum of this visual information represent?

This is the same set of presidential portraits, just a different image averaging approach. Image averaging is a process of looking at two or more overlapping images and blending those pixels in some way. Most of my image averaging examples are overlapping pixels moving towards each other, blending, or sort of splitting the difference between them in terms of hue. This other one above, instead, is the maximum channel values for all non-transparent pixels. It’s a very different image and feels like a collage more than a profound blob. I like how much of the foliage from Barack Obama’s portrait shows through.

One last image to share. I love surfing through all the generative adversarial networks (GAN) that are out there. There’s a nice collection of GAN networks here (https://thisxdoesnotexist.com/) some of my favorites are the ‘This Cat Does Not Exist’ or ‘This Art Does Not Exist.’ 

GAN, in short, is s a Machine Learning process to generate images. For example, each time you refresh this page you will see a person and it will look like a real person but, as it turns out, this person does not exist. 

I was surfing through this repository of fake humans and saving images as I went.

This is 40 generated images of fake humans all combined into one.

Back to main art page.