Man in the Moon 1.1.1 is out. The free browser demo is now almost twice as long: eleven days on the lunar surface, the first fork in the route, and nights that change every time you play.
https://t.co/omTJGViqPo
It started as a nature documentary.
Fat Bear Summer Stage 1, Pine Hollow, is out now for $2.49. The snow is Russia, coming next. Buy once, get every stage.
https://t.co/KSN3bauAgf
The Earth is too close.
Man in the Moon is a small lunar horror game I made with Claude. This is night 16. The first eleven nights are a free browser demo; the full game is $2.99.
https://t.co/omTJGViYEW
AI-assisted game dev is not a shortcut. I once spent a whole day on a fishing mechanic: 12 commits, 7:48am to 5:12pm, just to get a line, a hole, and a landed fish hanging the right way. And that wasn't the first day on it.
Spent an hour writing out exactly how one mechanic should work. Every state, every timing, every weird edge case.
It came back close-ish and confident.
So that was my whole day, fixing it.
Not sure if being that detailed helped or just made it easier to see how wrong it was.
728 prompts over 282 hours is one decision every 23 minutes.
The time in between is the job: testing it, finding what broke, deciding what to ask for next. Six hours a day for 47 days, after a full-time job.
That's what made with Claude looks like here.
@oldstackjournal I’m not sure what the answer is, but I’ve spent the last two hours fixing a fishing mechanic in one of my games, so I certainly understand.
Man in the Moon, made with AI: 42 days, 91 hours, 479 prompts.
That's 479 decisions I had to make. What it looks like, what scares you, what gets cut.
28 days from first test scene to live on itch. A small horror game, and it took all of it.
Free demo: https://t.co/omTJGViYEW
Day 68 of the JRPG I'm making with AI. 552 hours, 9,836 prompts, 1,761 commits.
I spent the first 260 hours on a version, then started over in HD-2D. Then last week I redid the art.
Every one of those prompts was me deciding something.
If AI makes game dev easy, nobody told me.
My unnamed 3D platformer so far: 47 days, 282 hours, 728 prompts, 730 commits. Level 2 failed its playtest and I rebuilt it.
All that got me three levels. There are 24 planned.
AI is why one guy can make this. It isn't why it's easy.
If AI makes game dev easy, nobody told me.
My unnamed 3D platformer so far: 47 days, 282 hours, 728 prompts, 730 commits. Level 2 failed its playtest and I rebuilt it.
All that got me three levels. There are 24 planned.
AI is why one guy can make this. It isn't why it's easy.
If AI makes game dev easy, nobody told me.
My unnamed 3D platformer so far: 47 days, 282 hours, 728 prompts, 730 commits. Level 2 failed its playtest and I rebuilt it.
All that got me three levels. There are 24 planned.
AI is why one guy can make this. It isn't why it's easy.
728 prompts over 282 hours is one decision every 23 minutes. The time in between is the job: testing it, finding what broke, deciding what to ask for next. Six hours a day for 47 days, and it's still nowhere near done.