Text-to-Image @ Adformatie
Featured in Adformatie’s October 2022 piece on the text-to-image hype, alongside six other creatives each sharing their take (and their AI-generated images) on the tech. My contribution: some AI illustrations and my hot (lukewarm, really) take on where this was headed.
Midjourney gets used mostly for photorealistic output, so I went looking for other visual languages I like: infographics, collages, semi-abstract editorial illustration. I was curious what happens when you feed the model abstract or scientific concepts instead of objects, to see how it “understands” them. No post-processing beyond upscaling and cleaning up glitches. Generating images this way is a different mode of making. Where Photoshop lets you build a vision layer by layer, here you have to picture the end result first and then put it into words: a kind of reverse engineering from output to input, with less control, and an invitation to embrace happy accidents. A tool like Midjourney creates hype less because of the technique than because it’s so accessible that everyone can join in. I see image generation mainly as a sketchbook on steroids. Change always threatens the status quo, the way photography once transformed painting. AI changes what it means to be a maker, but not the value of what a maker brings: intention, vision, taste still can’t be automated. What’s genuinely unresolved is copyright, given how little transparency or legal protection exists for makers right now.
