Generative Design: Produce more, by doing less.

I read something depressing at the weekend:  Medeival peasants worked less than modern-day workers. Whilst they may not have been materially rich, they certainly had A LOT of leisure time compared to us. But this isn’t about historical comparisons, or how we were lied to in the 1950s when we were told that technology was going to usher in an age of leisure.

This is a story about how I realised I could leverage creative coding to do something for me that I like doing the least… social media posts.

So let’s begin.

Like all great work ideas, which are generated whilst not at work… I was playing around with a design for a new painting. I like to sketch out the composition digitally before I render it indelible in paint because I am incredibly indecisive. Digital composition usually takes me an inordinate amount of time because I tend to get very fussy about colours. I can end up just accepting the last version because of the inevitable fatigue of realising that I have to choose something. A bit like trying to find a restaurant on holiday.

Being the lazy person that I am, I believed I could outsource this indecision to a generative script that could do the bulk of the colour picking and compositioning for me. This would create a huge data set that I could then curate and hopefully take inspiration from.

So I fired up the P5JS editor and coded up the basic composition, but added random colour generators, position translates and recurring shape loops. It didn’t take that long, P5JS is an intuitive library that makes creating objects on a digital canvas really straightforward.

Now that I had a data set of hundreds of potential ideas I could peruse, something that would have previously taken me a while to come up with manually. One thing I learned from the new data set was that I much preferred the chair to be centred, rather than on the right of the design as I had initially composed it. I also realised I liked bigger circles more than smaller ones. The main issue I still have is that the colours are just too random to make sensible use of. I will have to think of a way to develop more structured colour ranges but, for a quick and dirty test, I’m really impressed by what I’ve learned.

Then work mode kicked in, maybe I can apply this to the work I actually get paid for…

So as proof of concept I whipped up this little script that takes a heading, subheading and CTA. The background image is plugged into Unsplash API to pull a random image based on a keyword.

Very basic layout, but when you’re in the early prototyping stages with messaging with clients, it does help to give them some visual examples to stimulate discussion of where to take the campaign. When your colleagues want to move quickly initialising an idea, but you’re head down doing something else, it is difficult to pivot and put your brain into a different mode and come up with sketch designs to help them out. This (sort of) makes life easier because you can hand this to them and say “go nuts”. Come back when you want it refined to do something else.

Once you’ve worked out the template they want to use, they have an infinite resource to generate their own content without having to bother the designer with a irritating message change. What is even better is you have created a tool for them to use so they can do their job faster and get back to scrolling on Tik Tok.