Bebe Baker: A Postmortem

Final thoughts on Bebe Baker

This article is just my recollection of the whole project and my final thoughts on it. It was a long journey, all things considered, but it was worth it. It was never a commercial project from the beginning, so there’s no use judging it on those terms. The goal was to finish the game and ship it to different platforms, and I think I succeeded in that.

Where the idea came from

The idea started with a CodeMonkey tutorial. My first concept was a simple farm game where the counters would be animal pens — you’d deliver food and feed them — but that started to feel complicated once milk and health and everything else crept in. A bakery felt better, and I like bakeries a lot. So it started there. The code was already there, at least something, and I began working on it with the help of Claude AI.

Things went smoothly since the code was already great, but as I introduced new features, the bugs piled up — and the AI did an amazing job pointing out where the issues were. So as far as the programming side goes: always use logs, no matter whether you work in Unity, Unreal Engine, or Godot. Logs will save you time. In the beginning it might feel like a waste, but as the project grows and the lines of code and complexity increase, it’s amazing how many hours you’ll save yourself from hunting tiny bugs.

3D art

As far as 3D art and meshes go, it was a weird place to be. The game wasn’t large and it only needed a couple of assets. But on the online stores — including Unity and Fab — I couldn’t find what I was looking for, mostly because I didn’t know exactly what I’d need. The oven was a no-brainer, and so was the counter, but other stuff like flour and the rest of the ingredients came later in the project.

I decided to use AI from the start, knowing that many people don’t like it and disapprove of it — but again, my goal was something different. Now that everything is done, I’d still do the same thing: testing, playtesting, brainstorming, building. But at the end I’d swap in human-made assets. Why? Because by then you know exactly what you need, and getting those on Fiverr shouldn’t be too hard. The other thing is that pretty much every studio out there does this — just with far better organization and discipline.

I used Meshy AI, but even then you’ll probably have to drop into Blender to decimate the mesh (or Houdini, or whatever you use). The triangle and quad counts coming out of Meshy are really high, and their remesh tool isn’t great. This doesn’t mean a human-made 3D model is always the answer, but I anticipate a lot of these cases will fade as AI improves.

UI

The UI was a similar story. At the start I had a rough idea of what I wanted, sketched in Krita. It was terrible — just basic shapes. The first icons and images were made with Google’s Flow. The results were fine, but later, with GPT’s image update, I remade all of the images, though the icons still used Flow. I used an online background remover for those icons. I’m not sure if there’s a way to prompt Flow to do all of that, but this workflow worked better for me because I tested it and it was consistent.

I can’t draw, but even if I could, I don’t think I’d bother with these. If I could make them in a day or two, I would. But again, Fiverr seems like the better option as far as time goes. If you enjoy drawing — if it relaxes you — go for it.

Music and SFX

Music was something I looked forward to, since I can produce a little and work in FL Studio. But it’s mostly drum and bass or hip-hop beats with vocal sampling. When I tried to make something funky, or something fit for an adventure game, I flopped badly. So I decided to use Suno AI.

For SFX, I think the better option is ElevenLabs. They have short clips that sound realistic, and you can generate them fast and brainstorm with them — use them for inspiration. Each generation gives you a couple of these, so once you’ve finished everything you can swap them out for human-made ones. And if you’re the human making them, there’s no reason not to use them for inspiration to get going.

Marketing

Marketing was bad for me. I don’t know how to do it, I don’t have much money, and I don’t have a community with any real numbers behind it. If you have those as an indie, it’s going to serve you so well across social media and YouTube. My marketing budget was around $50, and I spent it on Google Ads and a TikTok video. That was all. But depending on what you’re posting, you might be better off with just images and screenshots on Facebook or Instagram.

If you use a lot of AI in development, the backlash is immediate. But all things considered, there’s a reason so many people work in the same studio doing different things. This was a great learning experience in marketing too. It’s hard to make content that’s catchy for gamers while doing everything alone.

Final words

Not much else to say. The show goes on — other projects are already on my mind. I’m developing a few, and I’ll try to improve in all the areas mentioned above. I like this, and I think I want to do it in the future. So maybe I’ll update Bebe Baker with more recipes, maybe I’ll swap out the meshes, UI, and music if there’s interest. But from this point on, I’m focusing on Metamorphosis, Scatterpoint, and finally Afterlife.

Till next time — see ya.