Comics with Gpt-Image-2

Playing with the latest generative image model from OpenAI on some personal comics.

GPT-Image 2 is definitely a solid, moderate leap in visual fidelity and overall quality compared to GPT-Image 1.5. One of the other things it really likes to do is never shut up: if you don’t explicitly prompt against it, it tends to err on the side of verbosity.

You can see that it really wants to show off its ability to embed text within an image, often placing way more text than you actually asked for.

I decided to throw gpt-image-2 at some of the comics that I’ve come up with over the years to see how well it would do.

Pac-Chart

Prompt: A slightly depressed looking Pac-man is shown strapped to a whiteboard (or something similar) with percentages on him to indicate that he has found work as a living “pie chart” (Parody of the old Flintstones line “It’s a living!”) in a board room filled with corporate executives which are the original Ghosts from the old arcade game.

Sisyphean Fruit

A comic showing a giraffe who grows up next to a small tree with a single piece of desirable fruit on it, each year the giraffe gets taller but the tree also grows a corresponding similar amount thus preventing the giraffe from ever being able to get the fruit

Canine Weirding Module

Prompt: A small dog has been bullied all his life by larger dogs. One day he discovers after having a minor surgical procedure and forced to wear the humiliating medical cone that barking while wearing it amplifies his voice in a Dune-movie sonic energy blast