Lady Bug Remake in Godot 4

Showcase Games
, , ,
1

ladybug_20260511_210512_956
ladybug_20260511_210512_956800×880 5.27 KB

Hi everyone,

I’m happy to share the first public release of my personal Lady Bug remake made with Godot 4 and C#.

When I was a kid, I really loved the original Lady Bug arcade game. I always found it had a special charm: the moving gates, the colorful maze, the pressure from the enemies, and the whole rhythm of the game made it feel different from other maze games of that era.

My goal with this project was to recreate the game in a modern environment using Godot, while trying to stay as faithful as possible to the original arcade behavior. A big part of the work was reverse engineering the original game logic, especially player movement, enemy behavior, rotating gates, collectibles, scoring, level progression, and the transition screens.

Another important part of the project was experimenting with AI-assisted development. I used AI heavily as a coding and reverse-engineering partner: to analyze behavior, compare implementation ideas, generate C# code, refactor systems, document findings, and iterate on the Godot implementation. It was not a “one prompt and the game is done” kind of process at all, but more of a long back-and-forth workflow combining observation, testing, reverse engineering, and implementation.

The project now has a playable v1.0.2 release for Windows.

Download page on itch.io : Lady Bug Remake on itch.io
Source code and GitHub releases : Lady Bug Remake on GitHub

This is a personal, non-commercial fan project made for learning, preservation, and technical exploration. It is not affiliated with or endorsed by the original creators or rights holders.

I’d be happy to hear feedback from the Godot community.

8 Likes
2

Very nice…

1 Like
3

It’s a very faithful recreation :slight_smile:

Did you have access to the original enemy behavior algorithm? They seem spot on, but I haven’t played the original much.

1 Like
4

I didn’t have access to the original enemy algorithm. I took the ROM and used Ghidra together with MAME to disassemble and analyze the code. But since I had absolutely no knowledge of the Z80 CPU, I used AI to help me understand the assembly.

However, a purely “static” analysis of the code is not enough. You also need to use MAME execution traces and debugger tools. For this remake, I probably spent around 80% of the time doing reverse engineering with the help of AI.

AI is a great assistant, but when the problem becomes very complex, you really need to guide it step by step. It’s a bit like driving a horse cart on a chaotic road: you have to hold the reins tightly, otherwise the cart immediately goes off the road.

For the enemies, I even created a separate project just to simulate and analyze their movement:

But it was still extremely complicated. At first, the AI analyzed the logs too heuristically, without relying enough on the actual source code. Then, when I asked it to focus primarily on the code, it wanted to simulate almost every assembly instruction one by one. The hardest part was finding the right balance between those two extremes.

I can’t claim that the enemy movement is 100% identical to the original game, but I think it’s very close.

This is the first time I’ve mainly used AI for development. I used ChatGPT with the Thinking mode.

I know many developers are now using autonomous agents, but in the case of a video game remake, I remain fairly skeptical. I think the process needs to be extremely iterative in order to make real progress.

For example, at first the AI kept making mistakes because it hadn’t realized that the graphics were rendered column by column, from bottom to top. To figure that out, I had to manually use the MAME debugger and understand what was actually happening myself. I doubt an autonomous agent would have been able to discover that on its own.

1 Like
5

I also use AI for code purposes, it really is a fantastic help for some things… good prompts in, good code out, is what I’ve found :slight_smile:

1 Like