Calyros DevLog #1 - Creating a heightmap and unit testing

Hey you, and welcome to my first DevLog for the so-far unnamed multiplayer RPG I'm creating.

This is actually the fourth post I've written, but in the first three I spent a lot of time waffling about what I'm building and why - stuff you might not care about.

I thought a better introduction to this project would be to dive in to what I've been building.

Writing tests

When I started the project, I began by writing tests.

During my day job, I'm a software developer, so when I'm writing code I'm not very comfortable with, I know that writing tests first is a good approach.

The point of tests (for the non-coders) is to provide a way for you to know if your changes have broken anything. When you modify the code - if you've written them well - running your tests will tell you if you've broken anything.

In my case though, I'm just starting with Godot. Unfamiliarity meant that I was breaking things more often than not and I was spending way more time writing and fixing tests than I was learning about Godot's features.

So I stopped writing tests. I want to go back to them in the future, because it's a good way to ensure you deliver a robust product, but for now, they're getting the way of progress.

Creating a heightmap

After abandoning testing, I moved on to my next task.

Whatever else my game may or may not contain, I knew it needed a map. By this point I was pretty sure I wanted a point-and-click tile-based game and so generating the map should be simple, theoretically. I just need a grid of 1x1 squares.

Step one was to decide on the size of the game world I was going to generate. I feel fairly settled on a vaguely ancient-Greek themed world, so I opened up Google Maps and checked the size of the area I wanted to replicate. Greece is around 300 square kilometers.

Even now, I knew 300,000 kilometers is probably too many kilometers.

Like the fool I am, I divided it by 10 and thought "that's a much more realistic world size" and got to work on generating 30,000 tiles.

The concept is pretty simple. Moving from right to left, and from top to bottom in a grid, I needed to generate a "height" for each point.

The height of Olympus - the tallest mountain in Greece - is about 2,900 meters. I rounded it to 3,000. Since I divided the world size by 10, I could just divide the mountain size by 10 too. So the highest point on my map would be 300m. For simplicity, I decided the lowest point on my map would be -300m. That would mean a value of 0 and above were land and below 0 was sea.

But I wasn't generating 30,000 data points, was I? That would only be the width. For it to be square, I needed to generate 30,000 by 30,000 meters.

Which is... checks maths ...900 million data points, oof.

Even if I was able to store the each point on the map as a single byte - which would mean the minimum variation in height between two points would be 2.5m - I'd need to store 858MiB. Okay. So 30,000 by 30,000 was unrealistic.

So I looked up some game world sizes.

• Skyrim is 24 square kilometers (really? I can run across it in 10 minutes).

• World of Warcraft is supposed to be 142 square kilometers.

• GTA V is 75 square km.

30 square kilometers was probably too many. But I noticed two things...

1. It doesn't matter how big your world is. Skyrim is the smallest of those examples, but it's packed with things to do and so it feels large.

2. Secondly, I'm not Bethesda or Rockstar. I couldn't make a world of that size feel alive if I wanted to.

So, I went back to the drawing board. I settled on 4 square kilometers this time, and got to work on my procedural generation code.

Now, as it turns out, using noise to generate 16 million data points takes a really long time. Especially if - like me - you're trying to do it in GDScript on the CPU.

I spent weeks doing this. Generate map. Write a tweak. Generate map. Write a tweak.

Finally, frustrated that I hadn't made any progress in the things that mattered and gave up, and I learnt my first lesson of amateur game development. Start small.

I moved the code I had written out of my project and settled on a 16 by 16 tile map. Small and achieveable.

Instead of trying to craft a whole world at once, I'm going to focus on building all of my functionality and features, and putting them in a small space - which I'm calling the "Pub Garden".

The time wasn't totally wasted though.

Lessons learnt

• Start small, focus on quality
• Procedural generation is hard
• Unit testing isn't going to work for me yet

Onwards and upwards

All of that time playing with procedural generation hadn't left me empty handed. I had generated and saved to disk a 16x16 meter heightmap. It had an altitude of about 10 meters and slightly bumpy "terrain".

Now that I had that heightmap though, I could turn my ambitions towards turning that into terrain.

For now though, I'll leave this post here.

~ Sam