Posted October 04, 2025 by Roboticy3
#update
Hello! Making a videogame while being a full-time student is no joke. I loaded up my schedule with some extra spicy computer magic this time, too, so I thought I wouldn't have time to update this game. But a couple small fixes here and there really help.
1. Controls can now be customized from the tutorial screen. One profile is kept for keyboard and controller.
2. Minimal controller support added.
3. Roach now controls better on moving platforms.
4. Roach no longer clips through waterslides.
There's lots of stuff I didn't have time to talk about for the jam project. This was meant to be a final sendoff for this project, but here's some things I didn't get around to.
I wasn't able to port the demo to the version of Godot that I was working with for this project, but older versions of RoachGame had npcs moving on the same logic as the player character. This meant that npcs could climb walls and follow you. I had an surprisingly smart approach to the pathfinding, where instead of using some complicated solver to get npcs to climb towards you, they simply aimed for a position in the world constantly orbiting around the player. This added "error" made them surprisingly resilient to unpredictable terrain. Since their true target was constantly changing, they naturally tried multiple things to get unstuck, and since the wall climbing logic is extremely flexible, they usually found something by pure chance. The way to get them stuck reliably was to get them to follow you onto a wide ceiling, then jump onto a floor below, leaving them circling your position on the ceiling forever.
The circling behavior also made them have a similar energy to excited dogs, which is always good.
Besides porting issues, visualizing the orbit was difficult. The orbit code was one of the stinkier pieces of script I had written for the project in its heyday.
One thing I got a lot closer to including in the project was a story about how me and my friends found a 3D scanning app and were able to scan some real life objects into the game. These had become a staple of the test level for me, so I really wanted to include them. But, it turns out the scans were using *really* high resolution geometry. I tried and tried, but nothing I could do could fit the scans into an export small enough to get around itch.io's size limit for individual files. Eventually, I compressed the meshes in Blender, but that corrupted the UVs (that's the way the 2D texture aligns with the 3d model. If you mess it up, the whole model looks unintelligible), and I compressed the textures in Godot, but that did all sorts of bad things to them.
Looking back, I probably could have done it if I wasn't rushing- compress the mesh in a way that preserves uvs and compress the texture outside of Godot's import window -but it was hard to keep a clear head. I had already recorded lines talking about the story, and didn't notice the issue until I ran test exports about halfway through the jam. So, I hurried, and ended up corrupting the scans.
There is a lot of concept art for the different areas and characters I wanted to be in the game. I meant to have a gallery for them in the developer commentary. The main issue with this idea was that the concept art is completely bonkers! I drew it myself and I can't make heads or tails of most of it. Big "came to me in a dream" energy. The most cohesive, and most ambitious, is this one here for a level that takes place on the backs of a moving flock of birds.
I'm currently writing this while slacking off on homework. This should be the last update to this project. I'm currently working on my next project. The estimated development time is quickly reaching the ballpark of years. So I'll try to make a simpler demo just for this winter. Maybe new years is a good deadline?