This was my first properly professional commercial game as a solo developer, and honestly, even though it was a relatively simple game, the production became a complete spaghetti-code mess.
The project got to the point where every time I made a code change, I had to restart the entire engine because the game simply wouldn't compile anymore.
On top of that, I massively over-scoped the game. I kept adding more and more "unnecessary" content, only to eventually cut most of it back and settle for a much simpler, more primitive version. Some of the content I had originally announced had to be implemented using an older version of the game that was less broken. This included the Hardcore Mode and local co-op.
I was honestly pretty sad that I couldn't deliver the full game I had originally envisioned. But, to be completely honest, I just couldn't fix the spaghetti-code mess this game had become.
And ever since its release, I've had a bit of regret about it. Even more so whenever I'd see a friend, YouTuber, or streamer play the game and run into bugs, sometimes bugs so severe that they literally couldn't finish the game because of physics issues.
After release, I tried to fix it myself. Unsuccessfully.
I even started rebuilding the game from scratch, but it was simply too much work. And I'll be real, the game didn't even make $600, so eventually I felt like it was a lost cause.
I'm not particularly proud to say this, but then I got Claude Code.
And finally, after a year and a half, the game not only got fixed...
IT GOT BETTER.
I try to use AI as little as possible in my development. For me, Claude Code is an assistant, not the person making my games. I like making things myself and keeping my projects very home-made. In fact, my current game and the two games I'm working on for the future are all built on an engine I created myself from scratch using C++ and OpenGL. I want to have full control over my technology and make sure a situation like this doesn't happen again.
But in this particular case, Claude Code was exactly the tool I needed to finally untangle this mess.
First, it fixed the compilation issue that had been making development painfully slow. Apparently, the culprit was a third-party asset I had used to communicate with Steam for achievements. I replaced it with the proper Steam SDK, and suddenly the whole development workflow became usable again.
Then came the physics.
Claude Code helped me discover that there was a project setting I had completely overlooked. For some reason, it was disabled by default, and apparently it's even deprecated in newer versions. Turning it on basically made the physics behave properly.
I don't know. Unity stuff, I guess.
But it didn't stop there.
The local co-op was completely fixed and is now fully playable with dynamic split-screen for 1-4 local players.
The Hardcore Mode was also completely fixed and improved, ending up significantly better than the original version I had made.
And then, as an extra, the game went from having only local co-op to supporting local + online co-op.
So now, for example, two players can play locally on the same PC while two more players join online.
On top of all that, the menus were completely remade, tons of new options were added, and proper controller prompts were implemented for different controllers using a Kenney asset pack.
The game now works properly with a much wider range of controllers, not just PC and Xbox controllers.
So, after a year and a half, I finally managed to give this game the update it deserved.
It's still the same game I released back then, but now it's finally much closer to the game I wanted it to be in the first place.
And honestly?
I'm really happy I didn't leave it buried forever.
Now I'll test this update and check if anything is broken or not working properly, so expect a new update of the 100% finished product.
This won't be the last Bald Ball update too, news soon.