Name the Danger dev log #2

As I progress on my development of the game, I was trying to figure out how to reliably build portraits for my characters. Dozens of them potentially. I started with vector graphics, but getting the proportions right was surprisingly difficult, and the 2D vector animations would limit their presentation to small viewports, never as full characters in a 3D map.

So I figured: what if I ask ChatGPT-Pro to draw different angles of the character I want, then I ask Claude Code to gather as many necessary measurements and data points from those angles, and build a 3D model from them that I could place in the world and even animate?

The subject was Pitch, a badger-folk sapper who spends his days either drinking or reliving a traumatic past in which he blew up innocent people. Sometimes he both drinks and relives at once.

As Claude Code traced the measurements on the photos and built the model on Blender, I thought, “there’s no way this is going to work, or else people would be doing it all the time.”

Ironically, it’s characteristic for this character to show up pantless. But Claude built the 3D character like this because I forgot to tell it to pick the photo version in which he’s clothed.

There are major issues with the textures: the details fall apart at the seams due to the textures coming from different angles. I’ll have to figure out how to fix that.

If you create a 3D character, you may as well animate him. That wasn’t working so well.

I believe Gen Z refer to this as mogmaxxing

In the overall sense of things not going fully right, I’d say this is working well.

Pictured: my face at these developments