Why do game studios create their own engines?

Bethesda is the example I’ll use but I can see this applying to EA, CDPR, & others. Bethesda has their own game engine and even created Papyrus for scripting. I don’t know the economics of game studios very well but I would imagine that the time/labor has to be huge to create a custom engine, documentation, and then to keep up with everyone all the time. Then there is the risk of falling behind in terms of features (RTX). It must also make hiring a pain. It must suck having to learn a whole code base at the same time you’re learning a new language. Bethesda says that they have hired modders specifically because they demonstrate their talent with Bethesda’s creation kit.

I have looked for this answer elsewhere and I generally read the same things: Its too expensive for a studio to pay fees to an Epic games because Epic wants a cut which will scale with revenue. I have also heard the “right tool for the job” argument.

The countless dollars spent on salaries, equipment, benefits, etc do not scale down when you release a title that doesn’t do well. Plus this means in some cases you end up in a situation like Bethesda is now, overhauling the engine so that you can even start Starfield & TES VI. More games more often = less distributed risk per title right? Why not ditch the game engine dev in favor of a more aggressive release schedule? Its not like a Bethesda would pay the same price I pay for the engine. They would get the ‘Bethesda discount.’

As for the “right tool for the job” argument: If you have an obscure platform you’re targeting like the NDS, then yeah maybe that makes sense. After seeing the success of fortnite/pubg or EFT, I don’t see why more companies don’t go this route. I mean why not have raytracing today rather than 2 years from now? You can write all of your custom stuff on top of Unreal. The framework is so modular.

Last point: If more game studios just decided to be the game studio people and they let the game engine people be the game engine people, then wouldn’t there be more incentive for a company like Amazon to make Lumberyard better? Wouldn’t more game engines enter the market?

Thanks in advance

1 Like

How many game engines you think are there?
All EA studios use the Frostbite. It’s free for EA. Also they have the money and the manpower to keep up with everything.
My guess is Ubisoft have a couple engines. One for Nvidia’s and one for AMD’s optimised games.
Bethesda have one engine alone and is having it for a decade…
I think Square enix (and all it’s subsidiaries) are using one or two engines.
Epic I despise as a company, so I despise everything they do, so I don’t care. I ignore them.

My point is, people and companies that can’t keep up don’t step up. They just license an engine.

It is only as complicated as you make it.
Most game engines use C#, JS or whatever… Programmers already know those. Game Maker Studio, an engine I have toyed for over a decade have moved to C#. Unity uses C# and JS, I think the Epic garbage company’s engine is also using C# for scripting… On top of that if your team created the engine they already know all the ins and outs of it. You don’t need to learn anything. If you license an engine you need to learn all the quirks and stuff…

I really hope they are making a new engine. Their old one dates a decade back and all they have done is polish it and add features on top of it. Their engine was a mess 2011 when it released. It’s a mess today. Fallout 76 is made on the same engine that made Skyrim. Still, modified and expanded but still a decade old.

The success of PUBG and whatever is not based on the engine at all… Hearthstone was made on Unity before Blizzard decided to use their own engine.

Pretty much because they decide what features are important to implement.
Games and my guess is GameEngines as well, are mostly modular. A new feature can just be programed and added to the skeleton of the game engine.
RTX is not a feature. Not yet. It is dust in your eyes to convince you to spend 2000$ on a graphics card. Lacking RTX support today is not a bad thing. Why not skip the RTX support in favor of a better designed games? Cyberpunk? Because money.
Game companies, especially the AAA ones, do not make games. I will say that again. The AAA companies aren’t making games. They are making money draining machines. Fortnight is a skeleton of a game dressed in super shiny clothing and everyone is going insane. OK, it’s fun, but it’s not a game. It’s a trap designed to drain money. Same as Overcrotch and hearthstone and all the other online multiplayer based garbage. On the other hand people are constantly complaining about unoptimised shallow crap… Cyberpunk sold 13million copies. That trash sold… nevermind.
My point is you don’t need the hottest new technology to make a good game.
Sorry. I kinda went on a rant…

I don’t understand the question… If more studios stop developing engines, how will there be more engines? Epic uses their engine to make their own garbage. EA uses Frosbite for their own games, Ubisoft uses their engines, Bethesda uses their own engines… The only people who license engines are small companies. They don’t really need to go for their own engine.

5 Likes

Yeah, your point that if everyone just used one engine, then all the engineers and devs would have a consoistent toolset, and be more productive.
I think it takes quite a bit to drag a company away from that model, and so the chunk of change they are “saving” by the DIY approach os pretty limited.

And then some big softwarehouses got a nice fat discount… But if they take too many years to release, and change engine without paying the exdtra…
The discount no longer applies… and the engine makers start the legal case (Starcitizen with Cry)

My two cents: You can’t direct development/are at the mercy of whoever built the engine.
Theoretical Example from a tools dev perspective: Let’s say we’ve been using a 3d content creation tool inhouse for a while and to make our lives easier we’ve been using something to control the renderer. If development of that something stalled and it wasn’t kept up to date with the latest version(s) of the dcc tool it becomes a cost/benefit calculation. “Is it cheaper to roll our own instead of having to fight with something broken? Yes? Let’s do it.”

Some companies only made engines until the games on top of them became more valuable.

Criterion, previously Criterion Game and Criterion Software before that, made RenderWare that was used by a large number of games back in its time including all 3 of the original 3D GTAs and a few of their side games too. They showed the engine off with Burnout, then were acquired by EA, Render Ware went in house, was torn apart and added chunk by chunk to EAs engines and then Criterion stopped existing pretty much.

It helps every now and then to have someone make their own engine. Games all start to take on the same feel after a while. Like source games are identifiable by just looking at them, others have more variety and occasionally a feature that makes hem different. Like I don’t think we would have portal today if we had only ever worked with Gold Source and never moved on.

And sometimes its hard as a game developer to convey to a separate company making and engine for you the exact features you want it to be able to do so itnis easier to just do it them selves.

1 Like

Creating an engine that fits what you want out of it, especially when trying to assure some longevity. The producer can also dictate what factors, to make the priorities on. Looking at what CDPR did, in their OG Red Engine [ca. '2011, which can still cripple newer rigs, should you crank errything upp]. Newer iterations of the engine, have refined some of demands, to a more scalable factor… All that, versus having to go fidgeting with an existing one, while avoiding straight up breaking it [if it would even be a plausible pairing, towards even a single game title release]. Also having to paying outside licensing fees/etc., to an engine producer [I’d rather see a gaming company, have everything handled in-house, over having multiple outside parties in loop].

2 Likes

Since Frostbite was brought up:
EA standardising on Frostbite has lead them to a lot of trouble, Dragon Age: Inquisition, Mass Effect: Andromeda, as well as Anthem had major issues because of the standardisation on Frostbite, which was an engine built for shooters, not for large open worlds, not to mention it had no provisions for RPG mechanics, or dialogues, so all that had to be bolted on to boot.

So not only are you working with something built by others, you’re also trying to extend it to make it do what you need it to do while not having access to people that know the internals of the engine (the Frostbite team was notoriously overbooked and unreachable, according to Schreier’s documentary, both for Anthem as well as for Andromeda). Usually this leads to performance issues as things are bolted on sub-optimally. Not to mention the risk of bugs when messing with badly documented stuff you don’t really understand.

If you’re in the business of building specific types of games it very much makes sense to take care of the engine as well. If BioWare had used their own technology, instead of having been forced onto Frostbite at least the technical side of that game wouldn’t have been such a meme as BioWare’s engine was proven to work and produce good results. Frostbite, over the course of three games, has proven to be a bad fit for RPGs.

One of the major reasons I feel SWToR failed (which does not use Frostbite, but a third party engine) as a MMO is because they didn’t control the engine, so they were entirely unable to iterate fast enough after the disastrous launch.
In contrast to that there’s Warframe, which has been using their own engine which is so well optimized for that specific game that it still runs well, and looks really good, on lower end hardware and has been doing so since early days. They did consider licensing the engine, but the concessions they would have had to make were deemed to not be worth the effort especially since they might even negatively affect Warframe, so they kept it in-house.

As for Bethesda (Game Studios), throwing their engine out would destroy decades of built up modding knowledge, given that mods are their lifeblood as many of their games are barely playable without, that would be an exceedingly bad move in my book.
Not to mention that, assuming they have people that are capable of working on that engine, learning an entirely new tool of such complexity might not end up being that much of a cost saver, but it will make you dependant on the whims of a third party if you buy/license an engine.

2 Likes

I thought that this was a very thoughtful answer and I got most of what I wanted to know, but I do have some follow ups.

Sorry I was specifically referencing Bethesda and Papyrus here. I should have elaborated. My intention was to point out very few know Papyrus outside of Bethesda & its modding community.

I don’t have experience working at a game studio so I didn’t realize the same people working on the render pipeline were also doing level design etc. My understanding was that these were entirely separate departments and everyone on the team would still have to read the documentation. I assumed that if you’re building all of the shaders, there is a learning curve if they decide to totally change the render in a specific way.

I guess I meant if a whale like Bethesda moved to the licensed model wouldn’t Microsoft or Google with infinite $ be like “man we need to make a game engine.” I would point to premium coffee as the example. Coffee was cheap everywhere and Starbucks proved people would pay more for better coffee. So, the premium coffee market was born.

1 Like

Unreal literally markets this exact thing as being a feature of building it from source.

I didn’t assume the big players would just roll with the release of whatever Epic or Crytek want to do. I figured they would just write all of their custom stuff on top of it. For instance- Bethesda’s ‘radiant ai’ could easily be a component that they add in on top of it. In terms of graphical fidelity and platform targeting every console cycle, wouldn’t it just be better to switch from UE4 to UE5 and start making the game rather than overhauling the engine? Also consider that this could work against you and delay your game trying to get the systems you want working properly. Bethesda has loads of bugs in every game they release. Same with CDPR.

1 Like

They don’t really. But on the other hand, if the guys who made the engine are in the office right over there. If question arises you don’t deal with tech support and email correspondence. The guys are right there.
My guess is there are people that overlap working both on the engine and on games.

To be fair I have absolutely no idea how the Bethesda things work.

Google can’t even keep their own game stuff in order. You are practically asking for one giant player to leave the race so another giant player can come in.

I have not tried Starbucks coffee, but from what I have heard I can’t really call it premium. Everything that is not espresso can’t be premium. Schwartz coffee ain’t coffee. It’s washed socks…

You are spot on about the importance of modding and I hadn’t considered that. That makes perfect sense.

Bethesda would have to carry more weight than a small guy who gets no tech support right? Look at Riot. I bet a player of that size gets developers from Epic to help them implement everything. Its one of the first with Nvidia’s low latency thing. Correct me if I’m wrong but isn’t tech support an indie studio’s problem?

their has been 14 total war games and 32 expansions on 3 engines in 21 years more counting dev time of the first game starting some time in the 90’s. the first engine was quickly replaced by a much better one and the third engine is based on the second and has only had 64 bit added since 2009 with the newest unreleased game believed to also be on the same engine. while the settings, story, graphics, and ui have changed over the years, the game play mechanics in battles have only changed a bit with hero features being added which is just a single soldier battalion with a lot of abilitys instead of the few most units have and a lot of features being taken away. naval battles were made simpler and simpler and whenever a part of the game was bugged they just removed that part of the game in the next game. now their are no navy battles, no bridge fights, no full sized city/castle sieges, simplified limited city mechanics, single type of settlements, ect.
from a business perspective they put little effort into the first engine and a lot into the second then made the third mostly off the second and have done very little else since then other then reskin the same sheep over and over removing bits of the game instead of fixing problems. very low cost in high level programmer time with most of the work being map, artist, story, writers, ect and marketing.

at the time no other engine could support 2-3k individual units with animations and attacks and hitboxes and collision ect in the late 90’s early 2000’s and maybe they could have used a new engine in 2009 instead of making v2 but they didnt and reworking the engine later was less of a investment and it shows when bugs form v2 were still in v3 and it mostly just added dx 11. in the early 2000’s when they started working on engine 2 unreal was still pretty new and even if it could have work(idk) tw 2 definitely worked and theirs some risk in business is changing from something that works to something unproven especially if it fails and people start looking for someone to blaim.

now from a players perspective i see total war as a dead to me since they gave up on history and only do this bs heroes shit like warhammer. i dont want to send a entire army agasint one hero and watch the one hero kill the whole army it feels like a dumb manga aimed at middleschoolers “its over 9000”. their is a lot less strategy in “send in superman and hit supersayan every 30 seconds” and its not enjoyable especially after so many features i liked got removed or simplified over the years. however i have a co worker who likes the warhammer stuff and doesnt give two shits about history or strategy so their is a market.

long winded point is theirs a story to why every game became the way it is and choices were made along the way and while maybe with your knowledge from this exact point you could look back and say “this is more efficient” maybe when game director so and so had to make that choice it was less clear.

A lot.

Epic owns unreal

We have Evolution engine from DE

There is the Unity engine out there.

Square enix has like 25 different in-house engines.

It all really depends on the game being made and if the engine being created is there to solve a problem or if it is worth using a 3rd party.

RPG Maker technically counts as an engine.

1 Like

There is a text adventure game engine…
Game Maker Studio is here…
There are like 5 other low key 2D hobby engines. There are also a lot of low quality 3D engines…

Someone tried to make a logical engine where you type sentences and the engine builds the game around it…

I’d assume that most of the consultants Epic sends out to help people implement new features don’t really know the internals of UE. They’ve been trained on the new feature (and one might hope, general UE usage), so depending on how good they are you indeed have a leg up compared to companies that don’t get this level of support and need to figure everything out based on the documentation (which, from what I’ve seen, isn’t all that great).

So yeah, in that case you’re better off than the small guys, but it’s nowhere near as good as having the people who build the engine sit in the next room…

I imagine that’s why Warframe’s engine looks so good (as in, appears to work really well, despite supposedly being a bunch of spaghetti), the people building the stuff and the ones implementing it into the game, if they are not the same person, likely have direct access to each other.
Meaning that it’s also a two-way street, unlike when using a 3rd party engine: if the “engine dev” sees the “game dev” struggle with “feature x”, they can just go back and change the feature. Or if the game dev needs a specific feature, it’ll just get added.

If UE lacks something you need. Well, it’s going to be a lot tougher to get that implemented, that’s for sure.

The hardest part, and biggest drawback, of building things in-house, once a studio gets to the size where building their own engine is viable at all, would be finding the talent, and retaining it.

1 Like

… if you’re good (and big enough to justify the outlay on maintenance) its an opportunity for competitive advantage.

i mean, id have traditionally been the 3d engine leaders, just look at the Doom/Doom Eternal engine.

it’s glorious. the id tech X engines are a massive asset to the company - and likely a large reason why bethesda purchased id.

another reason may be to work with other custom in-house tools they’ve developed over the years, to try new things, or simply to scratch an itch.

The number of engines is definitely on the decline though. like most things in software, it starts out with a huge number of independent implementations (when things are new, unknown and in an early stage of rapid development) and gradually the chaff gets filtered out until we have a few options that work well.

Back in the 90s, every game had its own 3d engine, pretty much. Things like directX didn’t exist - you had to write your own DOS VGA display driver for performance.

Times change.

I very much doubt any newer studio will write their own engine these days, they’d be starting from so far behind and having to spend massive amounts to catch up.

But those who have been making 3d software since the 90s (and Papyrus as mentioned above is one of those)… likely have their own legacy codebase that has evolved with the company’s requirements.

As to a company just throwing dollars to make their own engine - it doesn’t make sense to start from zero today. Even a company the size of Google or Microsoft with essentially infinite money will have difficulty finding the talent required to build such a project from scratch, and even if you were to find the next John Carmack for example, you’d be starting from 30 years behind.

You simply can’t buy 30 years of performance tuning, debugging, etc. for your shiny new engine.

Just license something that works.

So i guess to sum up: those who have their own are mostly carrying legacy code. Starting a new engine today, unless its some major paradigm shift like raytracing only or such - is a waste of time and money - and game development is expensive enough as it is now.

edit:
i’m talking about 3d engines specifically, as they’re the big budget high complexity stuff.

2 Likes

bethesda own id software, if they go making some engine from scratch they’re crazy. they’d surely adapt id tech6 or whatever version its up to now, for their own stuff, now they own it.

1 Like

I dunno, would idtech work for the style of game of Elder Scrolls. It does smaller areas and a different styles of game very well. I am not sure it would work.

Not sure, but I do know id were working on megatexture tech with id tech 5(?) to enable large open spaces with their engine (for rage, IIRC).

So maybe. If not the current version, an earlier one perhaps.

Yes that was for rage, and I never saw much come of it after, maybe they do use it but they did not say much about it if they do.