Archives For ue4

Game Engine Wars

March 22, 2014 — Leave a comment

GDC has bought with it the exciting news that many companies that make game engines are following in Unity and Game Makers footsteps in an effort to “democratize development” as Unity’s CEO David Helgason puts it. Most are following a subscription model with a variety of price points having been announced. This is great news for developers as access to professional quality features, cross-platform development and increased productivity just got a hell of a lot cheaper. I can get a fully featured game engine with full source for less than I spend on my internet connection per month. That’s pretty crazy.

The main downer I can see at the moment is most companies seem to be competing on cost rather than productivity. Wild eyed developers who see the shiny features and triple-A titles that the engines have powered are already announcing the death of Unity. In truth though the reason you buy an engine off the shelf is to save yourself time and money building, maintaining and improving one of your own. Most games companies are interested in making games not technology. As such cost is only really an issue if you end up spending more money than you save in increased productivity. Even at the indie scale we’re talking about such small number differences (tens of dollars a month) that even a mediocre productivity enhancement in a more expensive product will more than pay for itself many times over. Given the price point differences the cheaper end are really saying to me “look we know our product isn’t great but at least it’s cheap” only the cost saving isn’t radical enough to offset the potential loss in productivity between platforms. For example at this stage Crytek would have to pay me to use their engine over the competition. Their tool is just not as usable, which is not to say it’s a bad engine, because it isn’t, but the UX is just terrible.

Bad UX is a typical problem of internal tools. They are built in an ad-hoc manner to a “just good enough” standard and are often just plain terrible from the perspective of the user. Typically horrors involve choice paralysis, bugs and lack of a well designed workflow that matches users needs. Most companies survive this because they have a huge amount on institutional expertise to guide new users and new projects through the complex maze they inhabit. This knowledge also tends to be tribal, passed on by word of mouth and custom rather than by good documentation. As soon as you can’t rely on this institutional knowledge you are basically screwed and your productivity is going to be very bad until you learn it. This has historically tended to be a big problem facing teams adopting internal technology and is an even bigger issue if you are licensing someone elses technology.┬áThe reason Unity currently has far more users than other engine companies who have also had free offerings for some time is simply that their mission has been to empower developers so their product is just easier and better to work with.

I’d also be very skeptical about getting seduced by revenue sharing schemes. 5% gross revenue is an awful lot of money to ask and could well push a developer into making a loss rather than a profit on their game. For example you make a game in one year and sell $100,000 worth of copies. $30,000 of that will go to distribution (e.g. Steam), $5,000 goes straight to Epic (plus the $240 you spent per month) leaving only $65,000 to account for marketing, the creation of the game (you’re going to pay yourself right?) and any profit you want to actually make. For Unity the upfront costs are higher but that 5% cut is gone so rather than costing you $5,240 you are only spending $900. When your margins are slim that’s important.

A Wishlist

So what would I like to see from an ideal commercial game engine?

  • A company that puts empowering developers as their first priority. Without this the project is doomed. The entire mission for any new feature should be to make the UX as good as possible within the other constraints. The tools and code should be extensively documented and that documentation should be easy to search and use. Video tutorials and regular education opportunities would be excellent. Further support for the community is a must as very often people will solve each others problems.
  • A clean architecture for making games. This essentially means that the process of creating a new game entity should not be unnecessarily convoluted nor should the resulting entity be polluted with lots of unnecessary stuff. I’m very much biased towards engines that make use of an entity-component system these days. The architecture should make inter-entity communication simple and event driven. There should be templates for scenes and entities which provide good support for procedural generation. All the engine code should be well tested, by which I mean it’s covered by unit and integration tests further it should be easy to write and run these tests on game code.
  • The engine should be fast. Game developers still want to push hardware particularly on fixed platforms like consoles. This means at the very least the engine needs to be multi-threaded probably with a generic jobs system to make use of the cores that are available on various platforms.
  • The engine should be portable. If it can’t build to consoles, web, mobile platforms, Linux and desktop OS’s then that’s a bit of a deal breaker. Not because you necessarily want to deploy to all those platforms but because having the option means you can should your companies plans change. This is important from the perspective of productivity as porting a project can be an awful lot of work especially if you need to learn a new engine to do so!
  • Iteration times should be blazing fast. Data should be editable on the fly and it should be ludicrously easy to experiment with ideas. It’s even better if code can be live reloaded, usually from script to avoid the pitfalls of compilation where certain cases will force you to tear down everything.
  • It should be easily extensible. An editor you cannot customise is pretty useless. Doubly so if you want to fit it into any sort of Continuous Integration/Deployment or want to implement a more complex build process.
  • Integrated. At a minimum the editor should support various forms of source control. Support for other popular developer tools out of the box would be gravy.
  • Source access. Big, ambitious projects will almost always need to modify the engine internals.
  • Feature rich. This comes last. Shiny features are no good without the above. Unity has amply demonstrated you can have a feature rich editor simply by empowering developers.

Now that these opening salvos in a battle over license cost have been fired we’ll hopefully see a wider war waged on empowering developers with tools that really enhance their productivity.