Pros and Cons of Building a Custom Physics Engine

I am currently building a 2D platformer game called Platform Pixels. I chose to build the game using the libGDX framework because it can export to iOS, Android, and Desktop very easily. The framework also includes (optionally) the Box2D physics engine, but I chose not to use it. Instead, I wrote my own physics engine.

The purpose of this post is to explain why I chose to write my own engine and also detail whether or not I still think it was worth while. In order follow along, you’ll need to know what the game is like, so here is a short gif to give you a feel for the mechanics.

Platform Pixels Sample Gameplay

Now that you have some context around the game, let’s dive into physics engines.

Benefits and Pitfalls

To help outline the benefits and pitfalls of building a custom physics engine, I’ve centered them around four groups.

  1. Learning Experience
  2. Flexibility
  3. Performance
  4. Fun!

You might notice that these all sound like benefits but but don’t worry, the pitfalls are in there. Even though these four things may sound like benefits they might not be for everyone, and they all have trade-offs.

So, without any further ado, let’s get started.

1. Learning Experience

If you love learning new things and you’ve never built a physics engine before, I highly recommend giving it a try. Even if nothing tangible comes from it, I guarantee you won’t regret it.

learning

Not many people know this, but the creation of Platform Pixels was actually an accident. Before knowing anything about game development or even graphics programming, I was curious about the HTML5 canvas element so I started playing around with it. As soon as I drew a rectangle on the screen I was hooked. I wanted to take it further. This thinking eventually led me to create a very basic 2D engine.

So how did I go from drawing a rectangle to making a platformer? These are roughly the steps I went through to get there:

  1. draw a rectangle on the screen
  2. make the rectangle move across the screen
  3. make the rectangle bounce off the edges
  4. add gravity (downwards acceleration)
  5. add keyboard controls to make the rectangle move and jump

At this point, I had a rectangle that could move and jump around the screen. In order to take the demo to a full platformer, I just had to extend the collision detection to support arbitrarily sized rectangles with arbitrary coordinates. So I did that.

I won’t go into any more detail about how the engine works (that’s not the point of this post) but I will say that the process of figuring out how to build it has taught me A LOT so, even if Platform Pixels isn’t a success, I have already benefited from the experience immensely.

2. Flexibility

Flexibility is sort of an obvious benefit of building a custom physics engine. If you’re not relying on anyone else’s libraries, you can build whatever you want however you want.

In the screenshot at the top of the post you probably noticed how the character (Block Boy) can stick to walls. I also mentioned that the creation of the game was an accident. Well, the wall sticking was also an accident. This mechanic actually came from a bug in the engine when I first created it. I ended up liking it so much that I fixed it and implemented it properly. And, what seemed like a complicated feature ended up being just a couple lines of code!

Platform Pixels Wall Stick Code

Depending on the out-of-the-box physics engine you choose, adding a feature like this might take much more effort. You might spend a lot of time reading documentation on how to extend the engine, and it might also take a lot of code as to implement. but, since the engine I built was so simple and I knew exactly how it worked, this feature only took minutes to add.

Now, I should point out that not all custom engine features will be this easy. I’ve had to throw away more than one idea because I either didn’t know how to implement it properly, or I knew that the engine wouldn’t be able to support it because of an early design decision or existing optimization. I’ve also had to do one (thankfully only one) major engine refactor which took an entire weekend.

So don’t think that just because you build an engine you are going to get infinite flexibility. If you build it perfectly, you may get what you want. If you build it wrong, you are just putting up walls that your future self will have to break down.

3. Performance

Performance isn’t always a guaranteed benefit of building your own engine, and if you ignore performance then your engine will surely be choppy and slow.

Getting performance benefits really depends on your technical ability and the complexity of features that you want to add. The engine for Platform Pixels is very basic, so it allowed me to keep things simple and optimize aspects of it based on certain assumptions.

To give you an example, here are some of the assumptions I have made and the benefits that came from them.

  • ASSUMPTION: the engine will only need to support squares
    • BENEFIT: no complex math equations for polygon intersection
    • BENEFIT: collision detection is only about 50 lines of if statements (that’s it!)
    • BENEFIT: levels can be represented by PNG files so no need for a level editor
  • ASSUMPTION: all (almost all) objects in the world have deterministic positions based on play time
    • BENEFIT: support for very large levels because only objects in view need to be referenced
    • BENEFIT: object draw calls can take more time if they need

These benefits have been amazing. Unfortunately, there are trade-offs (as always). These assumptions have provided simplicity and performance but they have also limited the features that I can add down the road. For example, I can never add triangular spikes or round rolling enemies!

4. Fun!

Like performance, having fun building a custom engine might not be a benefit to everyone. Like I stated above, the reason I decided to build an engine was for the learning experience. As a result, I’ve had a lot of fun doing it. If you just want to build a game, then you probably won’t have fun writing your own engine.

I think there is one way to figure out if you will have fun building your own engine. Just ask yourself this. Are you building an engine because you want to make a game, or are you making a game because you want to build an engine? If you answer yes to the first question, don’t build an engine because you won’t be motivated to put the effort in to making it great. If you answer yes to the second question, stop reading this post, open your editor, and start building. No really, do it. Why are you still here?

Was it Worth it?

tl;dr YES

Now that I’ve covered the main benefits and drawbacks I’ve experienced while building the physics engine for Platform Pixels you must be wondering if I still think it’s worth it. Well, the short answer is YES, it was worth it. The long answer is a bit more complicated.

Since the major motivation for me was to learn something new, I really enjoyed the initial design and development of the engine. However, I’ve been losing motivation as time passes because I learn less and less as the game nears completion. Because of this, I probably won’t be building a custom physics engine for my next project unless I need something that I can’t get from an out-of-the-box solution. I don’t think learning will be a big enough motivator next time.

I’d love to hear your thoughts and opinions in the comments below! Thanks for reading :)

gamedev