AI-Generated Superhero

Variations of the superhero Quantum

In one of my previous posts, I talked about a mini-class assignment that brought life to the superhero inside me called Quantum. Here is the post if you haven’t read it 🙂

In creating Quantum I learned many interesting ways to design the character using various tools and some hard creative thinking. It took me about two hours to complete, but the poster came out just as I had imagined it, or perhaps even better.

But after reflecting on how time-consuming this was, I began to wonder if Artificial Intelligence could lend a hand to future projects. Can AI bring my visions into reality with just a brief text description? Let’s find out.

(P.S this is also another professor assignment 😁)

The results are in…

Well, this was interesting…right? Ok so is not exactly as I envisioned but it does come close to the original version I manually created on Canvas. The AI-generated images from Canvas on the left seem like the software tried to stitch together my text description by crunching different images together. To put it more bluntly it’s like it placed a superhero sticker on a postcard. Also, I said no cape! At least the green and blue costume looks dashing.

On the right, we have the variations generated by DALL*E 2 AI. I really enjoyed these two variations because the overall images are more concise and unified. They also capture a bit more of Quantum’s super ability to teleport between places using a quantum portal. However, the images look duller than in the original version and lack overall sharpness and detail.

I guess these AI-generated variations of Quantum weren’t too bad but they are not as precise as what I had envisioned. I wouldn’t depend on this software for more serious projects as it is yet not capable of providing sufficient detail in the image. However, I would still continue to play around with the idea as AI technology advances. Imagine being able to create images directly from your mind with just a few sentences. It’s like having a sketch artist, but faster. Could it be that detectives in the future will use a victim’s description to generate a suspect’s image using AI? You never know 🤷🏻‍♂️.