Happy Thanksgiving! We've had family visiting all week, which means lots of laughter, good food, and memory-making adventures (including a trip to Yosemite). It also means today's post is arriving a little later than usual.
Inspired by our recent travels, I decided to conduct a little AI experiment. My goal was twofold:
Capture the Feeling: Could I describe a memory in a way that would allow AI to generate an image that evoked the same feeling in me?
Compare and Contrast: How would the same prompt perform across different AI image generation models?
Putting Prompts to the Test
I chose two distinct memories from our Yosemite trip and crafted prompts to describe them:
Prompt #1: A Touch of Whimsy
"A winding road in a mountain neighborhood in the fall time with a dozen turkeys crossing the street."
Here's how the prompt fared across different models:






Prompt #2: Epic Grandeur
"Large granite cliffs that are almost 4,000 feet tall on each side, framing an epic waterfall in the center."
And here are the results for the second prompt:




Reflections on the Results
Looking at the images generated across both prompts, I have to say, Gemini's outputs were my personal favorites. They captured the essence of my memories most accurately, evoking a sense of nostalgia and warmth. Midjourney, on the other hand, excelled at creating a sense of epic grandeur, particularly with the waterfall scene.
This experiment highlighted the strengths and nuances of different AI models, reminding us that each has its own unique style and capabilities.
The Value of Personal Memories (and AI as a Powerful Backup)
Ultimately, while playing with AI image generation was a fun and insightful exercise, nothing can truly replace the personal photos I captured during our trip. Those snapshots are more than just pixels; they're imbued with the emotions, the shared laughter, and the unique perspective of that specific moment in time.
That said, this experiment made me realize the potential of AI to act as a powerful tool for helping to recreate memories when there are no photos. If I'm ever in a pinch, having failed to capture a photo of something, I now know that I can turn to AI models like Gemini or Midjourney to help me recreate the scene (which could definitely come in handy when trying to explain something to someone who wasn’t there).