From autofocus to AI image generation, convenience becomes a method. Control becomes conditional.
It only looks like its opposite. I work constantly, seven days a week if necessary, ten to twelve hours a day.. I work constantly, seven days a week if necessary, ten to twelve hours a day. Most office workers do not come close. I did not either, when I occasionally had such a job.
Convenience as a Starting Point
In the seventies, or perhaps the eighties, autofocus entered photography. I switched immediately. Convenience exists for a reason. Focusing is rarely the point. Light and composition are.
My colleagues disapproved. They looked down on my “noisy lens” that did the work for me. That position did not age well. Autofocus is no longer controversial. It is simply better. And now it is almost silent.
I am lazy in another sense. I have little patience for pens, brushes, and paper. I used to enjoy cross-hatching, though I should mention that I smoked weed while doing it. That habit is gone. The urge to make images is not. Cartoons, comic strips. Two years ago, AI made that practical again. That is how The Adventures of Connie Croma came into being.
Before that, there was AIZ, a magazine produced almost entirely from prompts, design aside. I have never minded being early. Or controversial. With new tools, hesitation is a form of irrelevance.
From Autofocus to AI Image Generation
As AI becomes more reliable, I return to making comics. Character consistency, once a persistent annoyance, is now largely under control.
Nano Banana and OpenAI appear to be competing in image generation. That tends to benefit those who actually use the tools.
At first, Nano Banana seemed slightly better at maintaining character consistency. That impression lasted about a week. Then the familiar decline set in. It reshapes images, turns rectangles into squares, squares into rectangles, and quietly consumes your credits. It performs adequately in Photoshop for removing small elements and for general graphic adjustments. Useful, but not particularly interesting. More for designers than for photographers.
OpenAI took a different approach. To my surprise, its image generator now operates in a way that is both powerful and fundamentally unstable. It is good enough for comic work. That much is clear. It also insists on generating full pages at once, four or five panels, complete with narrative and dialogue.
Where Control Breaks Down
That is where it begins to fail.
Change one panel, and the entire composition is regenerated. Not refined, but replaced. The result is often arbitrary. Sometimes incoherent. The narrative collapses. The panels lose their relationship to each other.
The explanation is simple. The system struggles to apply a local correction within a larger structure. So it starts over. As if a slightly intoxicated artist has taken control.
This is not unusual. It happens in human practice all the time. The difference is expectation. We demand more from AI than we do from ourselves. And often more than we demand from each other.
Laziness, in that sense, is not the problem. It is the method. The problem is what we expect it to control.