Create your own game engine but don't use it (2017)

I started in 2012 by making a 2D game as a university assignment using Allegro 5. Some time later, when I started game jamming, I decided to reuse some of its code. After a few games done this way I've separated the common code into an engine.

Now I'm at 36 games made with it, and I even have proper CI and containerized toolchains for cross-compiling on GNU/Linux for Windows, macOS, GNU/Linux (using either Steam Runtime or Flatpak), Android, Maemo 5 (Nokia N900), PureOS (Librem 5), Emscripten (asm.js/WebAssembly) and Steam Link. The engine also works fine on Raspberry Pi and FreeBSD (there's just no toolchains yet), and I'm looking forward to port it to iOS, DragonBox Pyra, Sailfish, Nintendo Switch and UWP (Windows 10/Xbox One), neither of which should be hard, I just need some time and/or devices. The only significant platform that would need some larger effort that I identified is Playstation.

Some of this stuff wouldn't be possible at all when using Unity or Unreal, or would have been much harder when using Godot.

Having said that, I probably wouldn't start writing a full-blown 3D engine if writing a 3D engine wasn't actually my end goal.

BTW. This "Lonely Cube" game could be easily written in a simple 2D engine using perspective transformations ;)

Ah yes.

I think I started my "game engine" as such a few years ago when I got interested in ECS and SDL and wanted to see what it would take to make a "simple game." It's been a constant, almost pathological, cycle of deletion, fence-painting and restarts. Fun, and educational, but I've come to terms with it likely never leading to anything productive.

But, even if it goes nowhere, it can be liberating to work outside the constraints of a framework, in a "chop wood, fetch water" kind of way.

Don't take the title too literally. The article talks about full featured 3D engine which took them year to build and it was still lacking features.

For 2D engine, it's actually a couple of weeks to build one. I have built my own in 2013 and have released 4 games using it so far, two of which were very successful:

1. http://sonofawitchgame.com/

2. https://store.steampowered.com/app/949790/Rogue_Bit/

I did not read the post, but the title reminded me of "Limit Theory". People have been waiting years for him to release the game, then in September he said he would release the source code, but sadly nothing happened ever since. I hope we will not have to wait years for that, too.

That being said, the game already looks amazing as it is, the graphics are phenomenal. I am pretty sure people would be willing to pick it up after he open sources it.

As far as I can tell, everything is written from scratch, in C++.

http://ltheory.com

https://www.kickstarter.com/projects/joshparnell/limit-theor...

I think there's a difference between "creating your own game engine and then building a game in it" and "creating your own game without a game engine."

The former is much less focused and you're much more likely to build things that are not exactly what you need, more general than you need, or just because it's some feature a game engine usually has.

As a counterpoint, I wrote a game engine, and proceeded to release 18 games over 9 years using it exclusively (http://www.mysterystudio.com)

I wrote a SDL based software renderer for the first game, then extracted some reusable code for the second game, and just keep using it and adding stuff.

It started as a Linux/Windows SDL thing. Later became OpenGL/DirectX. Then came Mac support. Near the end I added iOS and Android backends.

It was good fun. I'd probably not do it again today.

Mike Acton has some pretty common sense views on this: https://youtu.be/qWJpI2adCcs?t=2796

Basically if your making your own game engine then use the context in which you are working. If you are going to end up with something generic like Unity maybe you should just use something like Unity from the start.

"Create your own software X but don't use it".

The reasons provided - actually, one reason - is that whenever you need another feature (the second reason provided is the ability to run on another platform) it's easier to just use whatever is there already in the existing software rather than making that feature yourself.

This is a pretty general problem of writing your own parts or taking external pieces, often called "dependencies".

There are problems (and benefits) with those external pieces, not mentioned in the article - cost of learning (vs. cost of creating), cost of bugs (you usually know bugs in your code better), ability to fix bugs (sometimes external teams take forever to do that...), weight of features which you don't need but which exist anyway in the external software (that code takes space, time and sometimes require you to work with that a certain way when you'd maybe prefer another)...

I fully agree that educational value of creating such a system from scratch is quite significant though.

I started development of my own 3D-graphics engine in 2005 as I wanted to create a 'simple' 3D modeller, refusing to learn the industry packages. My first prototype was able to create convex hull meshes over half-edge geometry. Then another revision, starting from scratch, to implement a scene-graph. Another revision for undo-redo functionality and I was sitting already atop 40k loc. 14 years and 12 revisions later have turned my pet-project into a full blown 3D graphics package I am quite proud of. Development has slowed down ever since I became a father but the lesson learned is that developing a 3D-graphics engine (and a game engine ie logic later on) leaves no techique stone unturned. Would I do it again? Certainly, if one would rewind my age back to the day I was as h0rny as 10 apes together ;-)

Tangentially related, is there an open source game engine that's recommended for building games?

My experience in game development pretty much amount to an incomplete multiplayer tetris game.

Cross compatibility is a poor excuse to not use your own engine. You can use cross platform libs that give you exactly what you need, for example SDL for basic graphics or QT if you want GUI support.

Sadly, their game doesn’t seem to be available in my region on iOS.

I went to Blind Mystics and the website doesn't work.

Also rings true in “create your own php/JavaScript/python framework but don’t use it”