Exploring AI-Generated Imagery

The market for AI is rapidly expanding, and many industries have started integrating AI into their business models. Today, I decided to test a free AI image generator to see how accurate the results could be. I chose the topic, ‘Playing Frisbee with a cat at a park.’

Before using the image generator, I provided a detailed description of the image I wanted to create:

A young woman in her mid-20s and her cat are playing Frisbee in a park on a sunny day during summer. She wears a light blue shirt that suits her fair complexion, denim shorts, and white sneakers. Her brown hair is tied back in a ponytail. She has a graceful smile, reflecting her joy in the moment. The woman throws the Frisbee, and the cat leaps to catch it. The cat is a sleek and agile tabby with golden and brown fur. Its vivid green eyes are fixed on the mid-air Frisbee. The cat’s fur appears to ripple with motion. The cat wars a sleek black collar with a small silver bell. The Frisbee is a burst of color, featuring a lively combination of bright blue, neon pink, and yellow hues. The park is full of green grass, and sunlight shines through the trees.

I also sketched my idea before generating the images with AI to have a reference point.

Round 1
“A realistic and vibrant image of a young woman in her mid-20s and her cat playing with a frisbee in a park on a sunny day during summer.”

Image generated with Adobe Firefly

The generated image appears realistic, and it is set in a park. However, the woman is not playing Frisbee with the cat.

Round 2
“A realistic image of a young woman in her mid-20s throwing a neon pink frisbee, and her cat, a tabby with golden brown fur and green eyes, leaps to catch it. They are playing on a sunny day at the park.”

Image generated with Adobe Firefly

When I added more details to the prompt, the AI generator took a different direction. The image is no longer realistic, and what was supposed to be a neon pink Frisbee is now a ball. Moreover, the generator excessively used the neon pink color, such as flowers, sneakers, and the shirt.

Round 3
“A realistic image of a young woman in her mid-20s throwing a neon pink frisbee, and her cat, a tabby with golden brown fur and green eyes, leaps to catch it. She wears a light blue shirt, denim shorts, and white sneakers. They are playing on a sunny day at the park.”

Image generated with Adobe Firefly

Conclusion

This was an interesting exercise, but unfortunately, Adobe Firefly could not generate an image of ‘Playing Frisbee with a cat at a park’ as intended. The more I added details to the prompt, the more it deviated from reality, resulting in inaccurate and unrealistic outcomes, such as a woman with six fingers on her hand. While AI can be helpful in many daily applications, free image generation still has a long way to go before it can accurately produce specific ideas.

Leave a Reply

Your email address will not be published. Required fields are marked *