Generating New Yorker-Style Cartoons with AI
Harnessing the power of OpenAI’s ChatGPT to Transform Cartoon ideas into Reality
Ever wondered how an AI might perceive humor? As it turns out, not only can it understand a joke, but it can also “create” one! Let me share with you my journey of creating New Yorker-style cartoons with the help of OpenAI’s ChatGPT and Dalle-3 Beta.
For a long time, I’ve had this admiration (okay, slight obsession) with New Yorker cartoons. Their wit, their sharpness, their ability to capture an entire narrative in a single frame. And of course, I’ve had my own whirlwind of ideas brewing for these cartoons. There was just one small hiccup: my drawing skills weren’t quite up to par with my imagination.
Stable Diffusion & Midjourney
From my previous adventures with Stable Diffusion, it was only natural for me to attempt training some models and tinkering with existing ones to try to generate cartoons in that distinct New Yorker style. Unfortunately, the results fell short of my expectations.
I certainly gave Midjourney a shot as well. While it performed somewhat better, the results were still inconsistent, with a mix of hits and misses. The crux of the problem was the AI’s inability to grasp the details of the idea and the necessary composition for the cartoons. Characters would get jumbled, often being replicated multiple times, and other typical AI quirks would surface.
So I forgot about this idea for a while. Until recently.
Enter Dalle 3 Beta
While I was working with ChatGPT, a new feature caught my eye: Dalle 3 Beta. Without hesitation, I put it to the test, prompting it to conjure an image of a fluffy cat lounging on a sunlit windowsill during a lazy early summer evening. The results exceeded my expectations. What struck me the most was the AI’s keen understanding of the nuances in the prompt.
Recalling my cartoon aspirations, I felt compelled to experiment further. And to my delight, the very first attempts were nothing short of spectacular.
Moreover, the essence, style, and sentiment of the cartoons were spot on!
With some caption ideas already brewing in my mind, I turned to ChatGPT to generate more and perhaps stumble upon even better ones. This proved to be an incredibly fruitful decision! From there, all I had to do was pick the cream of the crop and give it a slight polish.
And just like that, within mere minutes, I had a New Yorker-style cartoon at my fingertips. Previously, with an illustrator involved, this would’ve taken me at least a day in the best-case scenario.
Note! The caption and signature, as well as some color correction and adjustments, were manually done in Photoshop. Dalle 3 Beta does have text generation capabilities. However, it’s not quite perfect just yet.
With the pipeline in place, I was inspired to produce a series of cartoons. The process turned out to be both effortless and enjoyable. In just a few hours, I casually generated 10 more cartoons throughout the day, all without a hitch.
Conclusion
Drawing the conclusion (pun very much intended) from this artistic journey: while questions surrounding artistic style and copyright undoubtedly remain, there’s no denying the revolutionary capabilities AI, and particularly tools like Dalle and ChatGPT, offer to artists. They act as a bridge, allowing one to traverse the gap between imagination and manifestation.
Personally, I’ve found immense joy in this process. It has granted me the privilege to craft within a realm I’ve always aspired to be a part of. The landscape of art and creativity is vast and ever-evolving, and with AI as a collaborative partner, the horizons are boundless.
P. S.
I’ve been so inspired by this cartooning adventure that I didn’t want to stop.
I’ve launched a subreddit dedicated to them.
NOTE! The name is inspired by this wonderful project.
I aim to post fresh ones at least once a week. And, if my work resonates with you, I’d be deeply grateful if you could spread the word! Thank you!
My other projects and links:
Or If you want to support my creative work, you can buy me a tea!
