Adrian Elton.
Independent creative Adrian Elton has worked on 150 days of ‘AI-Ad-A-Day', a personal challenge to utilise AI-imagery creation platforms and build advertisement concepts.
He began working on the series mid-2023, but started this challenge early January.
“I had no idea that it was going to be a collection of any kind at all, let alone that I'd be posting them on a daily basis,” Elton told AdNews.
“All I knew was that I was consumed with a genuine child-like wonder to bring ambitious visual ideas to life in a way that had previously only been the preserve of those with serious commercial photography budgets.”
Each ‘advertisement’, idea and headline are created by Elton and are done as an ‘educational, visual production tool.’
The advertisements are created using a range of AI software, including Adobe Firefly, Adobe Photoshop (BETA), EverArt, Google Gemini, Leonardo, Magnific, MidJourney, ON1 Photo RAW, Pimento, Reve, Sora and Topaz Gigapixel.
Elton said the aim of the series was to see how quickly he could bring his ad concepts to life without spending money on custom or stock photography.
“Despite being incredibly adept at retouching, there's a practical limit to what you can accomplish with stock photography," he said.
“As a result, there have been many strong concepts over the years that have ended up on the cutting room floor, as some things just can't be 'faked' by cobbling stock shots together.
“It's for this reason I was so excited by the potential this technology had to level the playing field in a way that had not been previously possible.”
The ‘advertisements’ are non-commissioned works but simply ideas that Elton had created, not attributing to any brand collaboration or official work.
Creation of each still image ranged from hours to several days depending on the level of redefining needed to be completed post-AI generation.
“Specifically, I wanted to see how far I could push the technology," he said.
“Would it do my bidding? Could it accurately interpret my most vivid visual intentions? And were there going to be any visual concepts that were entirely beyond its capabilities?
“There has been much gnashing of teeth and the way forward has often meant bringing together elements that have been generated separately, in ways that have called upon my retouching skills as a way of bridging the gaps where the AI has fallen short.
“Where I have explored what AI can do with a brief, ideas-wise, the results have been consistently underwhelming hackneyed cliches."
Elton has explored AI tools like DALL•E and MidJourney since they first emerged in mid 2022. He said that technology has evolved significantly.
"Having to originally set up Discord to access platforms like MidJourney also required a skill set that had more in common with the code-savvy programmer than the garden-variety creative," he said.
"The largest output you could generate was a square 1024 pixel by 1024 pixel, 72 dpi image. So not only were the images unlikely to resemble anything, but the results were so tiny they were unusable for hi-res purpose of print application.
"Nowadays, not only are the generated images frequently indistinguishable from real photography, they can be upscaled to large format print sizes using software like Topaz Gigapixel which can interpolate accurate photographic detail that wasn't there before."
AI-imagery software has developed significantly in the last few years, with the general public able to access high quality software with less experience or creative credentials.
"MidJourney required getting up to speed with a rarefied lexicon of technical syntax which kept out all but the most committed," Elton said.
"But fast forward to mid 2025 and almost anyone can play ball, prompting literally anything that occurs to them, in simple, straight-forward, unadorned language, using tools that are now hard baked into the software platforms and digital tools that they're already using on a daily basis.
"So while that definitely dilutes the value of imagery and those who have made it on a professional basis; for those of us who have both the aptitude, talent and experience; we find ourselves working with new tools that catapult us beyond our wildest dreams."
Elton said that there is no golden tool in AI-generation, where prompts and different platforms consistently lead to different results.
"Which of those results is going to be most suitable is dependent on what the objective of any particular project or concept is," he said.
“However, the best platform for prompt coherence (at this stage) is Open AI's Sora, Google Gemini and Reve for photo realism, Magnific for upscaling and Adobe Firefly for hyper localised edits.
“Consistency is the next great challenge - and commercially speaking, it's the one factor that remains the critical difference between parlour-grade novelty and industrial-grade game changer.
“It's very easy to generate a thereabouts image, but if it's replacing high-end photography, or ultra-accurate 3D models and renders, then it starts to become a potential legal liability if something is being advertised and the image exhibits various AI hallucinations.”
The next stage of AI-advertising creation is to prompt human models into various environments and contexts with accuracy where the requirement is broader than one hero image.
“If indistinguishable photo realism is the end game, then hyper-technical platforms like Comfy-UI that use Stable Diffusion and LoRAs, which are effectively plugins that help an AI model achieve an exacting level of consistency and realism," Elton said.
“But this tech is still far from being accessible at the consumer level.
"But as the tech is moving so fast, a lot of what was on the wish-list six months ago, are no longer pipe dreams. Which makes you wonder what the status-quo will be in another six months time?"
With what's likely to be a crowded market of creative production, Elton said that companies will succeed with great focused messaging and visually relevant material.
“We're going to see a deluge of AI-driven campaigns, primarily on account of the double-pronged magnetic forcefield that is exerted by the immense and unprecedented time and cost-savings," he said.
“Paired with the fact that micro businesses and brands are suddenly able to access levels of creative production that were previously only available to those with the deepest of production budgets, the market is about to get very crowded.
“The quality of creative is also going to plummet and suffer like never before as those without any prior creative experience, pedigree or insight will suddenly be able to cosplay the role of 'advertising' expert without having any underlying understanding of how branding, advertising, ideas or design work.
“The ensuing explosion of media noise is going to make it harder than ever for brands to cut through without deploying incredibly well considered, consistent and creative communications.
"Whether or not AI was used should ultimately be as irrelevant as what lens or tripod a photographer may have used in the good old pre-AI days.
“The ads either connect, and the messages land, or they don't. Great, focused messaging, expressed in the most visually relevant and intriguing way, is going to continue to win the day - however it’s made.”
Have something to say on this? Share your views in the comments section below. Or if you have a news story or tip-off, drop us a line at adnews@yaffa.com.au
Sign up to the AdNews newsletter, like us on Facebook or follow us on Twitter for breaking stories and campaigns throughout the day.