Prompting My Way to Better Visuals

For a project we're currently developing, I wanted to create a project management "planning style" personality quiz as an engagement activity at the beginning of a module. One of the things I've enjoyed most about using AI for visuals is how quickly I can iterate on an idea by refining the prompts rather than starting from scratch each time.

Step 1. Starting with a General Concept

I usually begin with a broad prompt and then refine it based on the results.

For this project, I started with:

"Create a 16:9 illustration with a construction theme to introduce a project planner personality quiz."

This prompt relied on an earlier prompt that generated the quiz concept itself:

"Create a project planner personality quiz with different personas. The quiz should involve 5 to 10 questions and make scoring simple, such as based on score ranges or whichever answer letter appears most often."

The first image captured the overall concept well, but it wasn't quite right.

Step 2. Revising and Refining

The initial illustration felt a little too youthful for my audience. It reminded me more of an animated children's style than something designed for adult learners entering the telecom workforce.

So my next prompt was: "Make this more photorealistic. Ensure all people are wearing appropriate eye protection."

 

While I liked the direction, I realized the rest of the module used illustrated graphics rather than realistic artwork. Instead of starting over, I continued refining the same image with another prompt: "Modify the characters so they are styled like a graphic novel for an adult audience."

The style now fit the audience much better, but the colors were darker than the rest of the course branding. My final refinement was: "Modify to a brighter pop art style. Ensure all workers are wearing proper PPE, with no jewelry and hair properly restrained."

The Final Result

Rather than using any one generated image exactly as produced, I treated each version as a design exploration. I combined elements from several iterations to create the final introductory slide for the activity.

One of the biggest takeaways for me has been that AI works well as an iterative design partner. Instead of trying to craft the perfect prompt up front, I find it more effective to start with a solid concept and then continue refining based on what I like and don't like in each version.