Revolutionizing Visual Content: How I used Artificial Intelligence to Replace Stock Photos
Have you seen these images?
You probably have seen them. If you're a designer, you may have to sheepishly admit to having used them as well.
Having worked in design for the last 12 years, I have had the opportunity to work in almost every touchpoint of the design industry. Illustration companies, big advertising agencies, boutique design houses and even big publishing houses have helped shape me into the designer I am today.
I currently work as a brand designer for Crux Intelligence. An AI-driven business intelligence platform that provides insights into your vast swathes of data in the most well-designed environment.
Working inside a business-to-business or B2B company has thrown its own set of challenges. The audience I have to target focuses on providing solutions to big problems and reaping monetary rewards.
Challenging the status quo of visual design is something I live for. I enjoy the process of pushing the boundaries. Not so much as breaking them but more so bending them to push people's perceptions of what is acceptable.
Having a different approach to tried and tested visual aesthetics does not mean shunning them. Rather it is to broaden the audience’s own boundaries to what they might be comfortable with.
One of the phrases my father uses all the time is “Don’t be a frog in a well” or as he says it in Hindi “Kooenh ka daddhu”. The phrase has its origins in Sanskrit, Japan, and China. The Japanese have put a poetic touch to it by saying “The frog in the well knows nothing of the sea”. Growing up I used that to challenge my own knowledge. Satisfied but never content. Because contentment leads to complacency. Complacency is the point where people lose their imagination for new things. For change. For things, they do not know about. For things, outside their own little wells.
Through this blog, I shall explore how I used artificial intelligence to challenge the status quo of stock imagery. And maybe how I hope it will bring a positive addition to creativity and design.
Why attack stock images?
The use of stock imagery has been vetted by big corporations and people in B2B businesses to convey an idea or a thought in the most cost-effective way possible.
But this comes with its own pitfalls.
Brands depending on stock imagery to get their ideas across fall into the depths of inauthenticity. Our viewer is not the viewer they were a week, a month or even a year ago. The digital landscape changes rapidly every day.
Our viewer is smart, tech-savvy, consuming large amounts of digital data, and surrounded by screens. Visual imagery is important for them to distinguish between different information being presented to them. Memorable imagery is what really stands out to our audience.
Stock imagery is a cliché.
Companies these days are focused on trying to generate content for their audience. The use of stock imagery supports that habit. The use of stock imagery comes off as inauthentic, generic and downright spammy. The urge to report/block emails using stock imagery is real. More real than the stock image. 🫢
Well, where are the solutions which cater to companies that use stock images? How do we convey teamwork without using that picture of business people in suits joining hands in the centre?
Enter Artificial Intelligence. A technology that has picked up a reputation all the while being programmed by its human operators. I shall be staying away from that conversation and maybe discussing it at length another day. For now, we shall focus on what role AI could play in helping brands create their own distinct visual style and aesthetic.
AI has made huge advancements since its inception in the 1950s. From expert models to machine learning, the strides AI has made in mimicking the human mind are extraordinary. Deep learning, image recognition, natural language processing and recommendation systems have paved the way for a more customised digital user experience.
Leaning into the idea of customisation, I explored AI to see if I could create images for our company blog that would be more on brand and help define a visual language for a company which works with AI itself.
I started exploring the two tools which were being used at the time Dall•E 2 and Midjourney.
Process
Choosing the Right AI tool
Early explorations with Midjourney required many prompts. The Discord server interface took a while to get used to and for some reason, Midjourney creates more wonderous illustrated images than ones you could use for editorial purposes.
For the sake of seeing the difference between the two, I will be using the same starting image as a prompt for the blog. The starting image was a stock image used to signify growth and what better way to do this than through a young plant in nurturing hands.
Midjourney gave me the option to put a reference image in and then use that as a stepping stone to generate the image.
After a lot of trial and error, I managed to figure out a process by which I would get personalized images from Mid-journey. But this process would require additional steps like taking the image into Photoshop and applying some effects to make the images feel like they are made by the same organization.
This led me to try Dall•E 2.
I was quite taken aback by how Dall•E managed to make almost a new stock image for me using simple prompts. The edit feature let me stay on the Dall•E website and not use any other software to get my desired result. Some images did require a lot of generations to get right or remove an imperfection. But with Dall•E, all those were easily handled as well.
You are limited by their ability to not engage in conversation but are also astounded by the gigantic bank of knowledge it can draw from.
I decided to use Dall•E for my AI endeavours, as I felt the quality of the images I was getting from my initial explorations was more in tune with the editorial nature of the blog posts of the company.
The results of the plant image generated by Dall•E will come in the next section.
Crafting a Prompt
The key to using AI I realized was figuring out which prompts produce primarily similar results and what to expect when using certain keywords in them.
I have always been fascinated by the futuristic look and feel of chrome as a texture and metal finish. Do its reflective properties seem to shift time?
The Prompt system in Dall•E takes a little getting used to. I managed to find the suffix to my prompts after a couple of tries.
My prefix would change according to what I want to show, and the suffix would make it the same style, look, and feel I wanted the organisation’s blog to have.
After many experiments, I managed to get the pink and blue I wanted which fit the brand colours and managed to make the images feel futuristic and stylized. The chrome-dipped objects just added that bit of imagination and elevated everything a bit more visually.
After managing to get the image I feel is the best I turn to edit the image in Dall•E.
Editing the Image
Editing the image in Dall•E Outpainting editor made sure I did not have to jump to photoshop to do any editing. I could erase spots on the image I wanted to replace.
I can use the ‘generation frame’ to get proportionately the size needed as well as add more elements. or just simply extending the background of my chosen image. No more photoshop required 😤
My final image pushed the boundaries of what even I had imagined.
The difference between where we started to where we had come from was apparent. AI was a breath of fresh air to the tried and tested stock images we had grown used to and dare I say… we were comfortable with.
Conclusion
AI as a tool needs to be used.
They should be used to make human choices in the process seem more deliberate. It does not blur the lines between the machines and ourselves but makes them feel more deliberate.
The reason why machines are invented is to increase efficiency. That's why we have progressed as designers from people cutting up pieces of paper to sitting at our desks. Which was better though?
AI tools allow us to tinker more. Explore more. Imagine more.
Maybe even love more.
The reason we use tools to increase our efficiency is to increase the time we have to savour life itself.
AI tools help us do the jobs we find mundane. the jobs we do not want to spend our time doing. The jobs we would otherwise make our interns do.
Through gritted teeth, find that picture of the perfect pair of hands over a keyboard to signify office.
Maybe it is time to redefine that. Maybe it's time to introduce people to new images. Images made by machines but with a human element. Restricting ourselves only through imagination and not by skill. Making the term ‘Gatekeeping’ almost obsolete.
We frogs are being given a lifeline. Something to pull us out of that complacent well we have been sitting in for so long. Maybe show us a new future.
Maybe it's time we frogs got out of that well and onto the beach. Maybe even get a starring role on Baywatch.
