Visuals to Words to Visuals

I've written before about turning words from a brief into visuals. But with all the prompting going on, what happens the other way around – when visuals turn to words turn back to visuals?

When I design something, I can see it in my head. So the idea of needing to take that visual and turn it into words to then turn it back into a visual always feels odd. Yes, I can describe the way I want something to look, but I could also just visually create it and skip a whole step. At its heart, AI is programming, and so the input requires you to write what is a piece of code that we happen to call a prompt. It’s specific language written in a specific way that a computer can understand. But ultimately it’s not THE thing in my head. And I think this is why more often than not, visual AI results fall short.

Does the act of committing a visual to words make it less good? Kelli Anderson wrote about this idea in her amazing popup book, Alphabet in Motion:

“When we jot down a thought, we trade a personal, firsthand, deeply embodied sensory experience for something more standardized. Once committed to typesetting, font selection, and publishing, the firsthand becomes material – our inner world becomes externalized as common property, and our ideas become fixed. Philosophers from Ludwig Wittgenstein to Bruno Latour have sought to differentiate the realms of reality that can and cannot be glimpsed through the written word, while generations of visual artists have sought to give tangible presence to human experiences that fall beyond language's edges.”

She goes on to make the important distinction between letters as content and letters as forms/typography in graphic design:

“Typography may very well be the precise location where this paradox is resolved. Letterform is simultaneously a verbal and a visual experience. A letter can be thought of as a codified symbol helping us communicate, but a letter is also a shape that we see. When we write letters – in sentences and paragraphs – we use them as a tool to package our ideas so we can share them. But also: When we examine a letterform as itself – think of a giant A in a type specimen – we engage in a different kind of viewing, one that is a sensory, aesthetic, cultural.”

Henrik Karlsson writes about this idea of when we should visualize and when we should write from a writer’s perspective:

“When we put words to a thought, we have to compress something that is like a web in our mind, filled with connections and associations going in all directions, turning that web into a sequential string of words; we have to compress what is high-dimensional into something low-dimensional. This has all sorts of advantages, which I will return to, but the point I want to emphasize here is that compression is effortful.” – Source 

This helps me understand how I visualize something  – which is not as a solidified thing, but as a continually shapeshifting thing that maintains connections to other things. Only when I need to describe it, which is of course often necessary, does it settle into something. But the possibilities attached to it are now severed in some ways. Whereas if I were to sit down to make the idea directly, those connections could still have some flexibility in that process. Again from Henrik Karlsson, "Writing forces precision, which can fool us into locking in details we have no reason to lock in..."

I think this is the key to why comps and moodboards can be so tricky. You need to make visual ideas concrete in enough of a way that others can understand – “turning a web into a sequential string” – without somehow fully committing, so that you can maintain some shapeshifting and flexible qualities.