3 weird obstacles & tips to overcome (Well, at least how to overcome the first 2.) Plus jump down for some GenAI imagery that just… blends, making it rather ineffective at capturing attention in an ad.
I’m not a graphic designer, nor have I ever claimed to be. I tell clients I’m not the one to make things look pretty. I don’t have a great sense of color. I wear makeup recommended to me by the pros (and now my teen daughter.)
But I know messaging, so with these new GenAI tools, I can take care of some image creation! We can get things done faster and cheaper, right?!
Ugh. Sorta.
Most of these obstacles are probably widely known by designers leveraging GenAI, but that’s not my world. When I encountered these obstacles, the experience was like seeing a chipmunk dart in front of my mountain bike – a total surprise – but much less adorable and fun.
1. Spelling
The first time I encountered spelling errors was ChatGPT. I type quickly, but not very accurately, so I looked to ensure my prompt was spelled correctly. It was.
Gemini does it as well. You type text into the prompt and it spits out your image with numerous bizarre spelling errors. I was absolutely cackling at the outcome of putting some extra i’s and t’s within the word “security.”) à SECURITITY.
This still surprises me the most, considering web searches are the opposite. Misspell in Google and it knows what you mean and automatically corrects. Cool.
Spell properly in many GenAI tools and it automatically misspells. Weird.
Solution à Never ask for text.
Give me an image to represent XYZ, then I’ll paste it into my graphic design tool (PowerPoint, stop laughing) and add the text myself.
2. Inability to Edit an Image
I’ve had Gemini provide images I liked, but wanted to make small tweaks. When prompting again, I get an entirely new image, often a substantial departure from the first one (which… I had liked.) Gemini informed me it was unable to edit an image it had created. I was working with Google Docs, so perhaps it varies.
3. Inability to Isolate Image Components
Gemini created an image with a coffee mug, a martini glass, and some text above. I really liked the line art style drawing of the mug and glass, so I asked it to give me those components alone. I thought to circumvent the misspellings and inability to edit, I’ll do it myself! I would drop those two images into PowerPoint and update the background, sizing, text, add our logo, etc.
Nope, still No-Can-Do. It can’t pull out and isolate the components of an image it had generated for me.
AI Imagery & Ads = Blending, Lost in the Crowd
How can you get attention looking like everyone else?
I tried LinkedIn Ads AI, which is named “Accelerate.” I’m chuckling considering how my friend and fellow tech marketer said AI can “Accelerate your ineptitude.”
LinkedIn Ads delivered the below image for cybersecurity. Visually, it’s well done. But it is BOOORRRING!
And it does absolutely nothing to convey our unique selling proposition or intrigue you to learn more.
It gave me another image like the above laptop, and another like the below phone + laptop, 4 image options, but only 2 concepts.
The below conveys cybersecurity, but specifically for a user to protect their own identity and data with tools like 2FA. That’s distinct from B2B cybersecurity.
But of course, those are the types of distinctions AI doesn’t understand.
Do you need content marketing support from a human who understands cybersecurity? Let’s chat.