How I use AI

Made with generative AI

MISPRINT is my game. I set its direction, make the design calls and decide what goes in. I use generative AI throughout its development, for the art, the music and sound, the code and the writing. Here's why, and how it fits together.

Why

MISPRINT is made by one person, in the time I can find, without the budget for a team. Generative AI is how a game this size gets made under those conditions.

If MISPRINT finds enough players, I'd love to bring in artists to make it better: illustration, music, sound. If you're an artist and that sounds interesting, say hello at hello@roguelite.tech.

Art

The illustrations, from the shop cat to the clients and the key art, begin as AI-generated images. Each one then goes through MISPRINT's print pipeline: it's separated into four spot inks, then halftoned, grained and set slightly off register as the game draws it. The generated image is the source material. The print is what you see.

Music and sound

The soundtrack and sound effects are made with AI audio tools, then chosen, trimmed and mixed into the game. That includes the trailer music.

Code and writing

The code is written with AI coding assistants, and so is much of the text, this page included. Everything gets read and played before it ships.

The decisions

The rules, the balance, and what makes the cut. AI tools make a lot of drafts. I decide which ones become MISPRINT.

While you play

Nothing is generated while you play. The game runs offline, and everything in it was made ahead of time.

Curious?

If you want to know how something was made, ask at hello@roguelite.tech or on Discord.