I certainly doubt that GSC will attempt to develop another proprietary engine.
— Me, April 5, 2010
That’s the kind of wisdoms you can expect from yours truly, my friends: completely inaccurate ones. Word comes across the vines of grapes that Ukrainian developer GSC Game World is hard at work on S.T.A.L.K.E.R. 2 with plans for a 2012 release, and will – contrary to all sanity, logic, and sense – be developing a new engine for it in-house. So it’s a good news/uh-oh sort of situation.
Let me tell you something about game engines: they’re the hardest, most complex, most time-consuming, most difficult aspects of development. Developing one costs millions of dollars. There’s a reason that a garden of middleware developers dedicated entirely to creating engines has sprung up, and why larger companies such as Valve and Crytek license their engine software. If you’re a game developer and you choose to create a proprietary engine for your game, you should expect to tack 36 months onto your development schedule.
Let me tell you something else about game engines: they can be millions of lines of code, they must play nice with a huge array of hardware combinations and operating systems, and – even if they’re technically sound – they must be tightly optimized so they run well on lower-end machines. This is not a trivial task. Many small developers that opt to create their own engines either go out of business before the game is even complete, or they release a game on an engine that’s buggy, poorly optimized, unstable, or otherwise insufficient for its purposes.
This is what GSC Game World did – the X-Ray Engine was developed for STALKER, and it was a mess. Updated for Clear Sky, it was still a mess, only this time it was a DirectX 10 mess, meaning that not only was it a mess, it was a 1 frame per second mess. By Call of Pripyat they’d cleaned up and optimized the engine, turning it into a good performer, though its frame rates were still lower than they should have been. It was around this time some began speculating that STALKER 2 might license the Cry3 technology from Crytek. Cry3 would be especially suitable for a STALKER game given its ability to render vast, open outdoor environments, and Crytek has promised that unlike Cry3’s predecessor, this version will run well on midrange computers.
Until this announcement: yay! More STALKER! Boo! Proprietary engine!
STALKER was delayed and delayed and delayed and delayed – and sort of gutted – because GSC, then a very small company, insisted on developing a proprietary engine and AI. Given how ambitious the game design itself was, the decision to develop so much technology in-house was a terrible one. It nearly cost the world the game. There was a time when it looked like STALKER wasn’t going to ship, and given how important the series has become, that would have been heartbreaking. That GSC would choose to make the same mistake twice (and it is a mistake; Gamebryo and Cry3 are both well-suited to STALKER, and neither is nearly as expensive as the cost of developing a dedicated engine) is either a sign of hubris or a sign that the company has expanded to the point where it’s supremely confident in its own technical abilities.
Is this decision going to keep me from buying the new STALKER? Certainly not, and it shouldn’t keep you from buying it either. It’s just a bad decision, one that’s likely to delay the game beyond its planned 2012 ship and may result in a buggy, underperforming result.
But hey, a new STALKER is a new STALKER, so now I have reason to live at least until 2012.
Send an email to the author of this post at steerpike@tap-repeatedly.com.
I think the best part of the first one was the disorienting and disarming engine bugs. Very early on I was treated to a bug where every single polygon was outlined in gray and as I tilted the viewpoint the polygon fill in effect was something. You just can’t go and buy that sort of thing from iD. Clearly, part of the resurgence in indie games, some with pretty retro or simpler stylized graphics, comes from teams wishing to spend their creative talents on something other than tech and that’s a good thing. GSC probably has the money now to do what they want… I wish them luck.
I didn’t play the third one. I sure hope they come up with something special to put in the game, because I think the series has outstayed its original topic a little bit.
Mad. No, interesting. No, mad. No inte- we can only hope that GSC know what they’re doing!
If the X-Ray engine was a mess to begin with then maybe each subsequent version of STALKER was merely ‘polishing a turd’ and so with a new engine and three games under their belt (the last of which was apparently their most consistent) perhaps they’ll be able to hatch an engine which holds together properly. I’d like to think there was a very good reason for taking another crack at it because shunning a bunch of well developed existing engines seems totally gaga to me. I’m betting that it must be something fundamental to STALKER 2 which we obviously don’t know about. Who knows? STALKER 2 might be what GSC originally intended Shadow of Chernobyl to be. We certainly didn’t get that game.
I’m with you regarding the subject matter getting a little long in the tooth Helmut but I’m intrigued to see where they take the series now. I’d actually be happy for them to push the survival aspect some more and create a more definite connection with the people you bump into along the way. That was something I thought was missing from STALKER and after briefly playing Metro 2033 I was fascinated by the station’s inhabitants and their fragile existences all boxed up underground.
The only developers permitted to make game engines should be id, Crytek and Blizzard. I see no reason why anyone in today’s Market would wish to build there own. As you said Steerpike, it must be a bloody nightmare!
Admitedly they do have a lot of experience now; however it still won’t change the difficulty, simply the ability to overcome the obstacles through resources and man power. Surely money and time would be better spent using a pre-existing engine, altered to your needs?
If they have a team capable enough of producing an engine from scratch, they have one good enough to alter a pre-existing.
On a side note- shit engine or not, Stalker still has the best A.I I’ve ever come across. So it wasn’t all wasted (or bad; well, when it worked) 🙂
I’d add Epic (Unreal) and Valve (Source) to that list, but otherwise I agree!
There’s no point in developers making engines. Id Software isn’t really a game developer; it’s an engine company that happens to demo its engines using games. Crytek is kind of similar.
For my money, the most powerful, versatile engine available today is Gamebryo. It’s been used for everything from Defense Grid to Fallout 3 to Civ IV to Warhammer Online. It’s only around $200,000 for a full commercial license, it includes the mighty Floodgate technology to take advantage of PS3’s bizarro hardware, and it integrates well with Havok. While Unreal probably comes with better support, it’s really only suited for shooters and gets grumpy when rendering large outdoor spaces. Plus it’s cost-effective to license middleware physics and AI like Havok and Kynapse rather than depend on what the engine uses. You can’t beat it.
STALKER’s AI is second to none. Hell, if GSC wanted to make their AI into a standalone product and license THAT, I’d be okay with it. Engines they should leave to the experts.
Crap.
Except, id won’t be licensing id Tech 5 to anyone except ZeniMax and Bethesda: http://www.eurogamer.net/articles/2010-08-12-id-tech-5-only-for-bethesda-titles
Which is dumb
I haven’t said anything on this as I feel like I’m not nearly as knowledgeable on the subject as some of you guys are. That said, I felt limiting the number of people developing game engines to a very short list could lead to stagnation and lack of innovation. Economists seem to constantly stress the importance of competition which would mean the more people creating engines the more innovation we could see.
Lokimotive’s comment seems to only reinforce the idea of why you don’t want a few people controlling so much stuff. Again though, I don’t know much about this subject..
@Armand: Economists would also stress the ability to get shit out the door and sell something =) Engines are huge undertakings and can have disasterous effects on developers who attempt them and it’s NIH syndrome writ large. There are quite a few engines out there already for anyone who wants to use them; the market is quite healthy.
Did anyone ever use GSC’s X-Ray Engine? Did they get any licensing revenue from it?
@Harbour Master: no, they did not license it and only crazy people would’ve taken them up on such an offer if they had. The X-Ray Engine was a mess. Technically it has a lot going for it: great DX10 implementation, huge outdoor environments, solid physics, and visuals on par with Source. But the renderer was always buggy and unstable, they couldn’t reduce load times to a tolerable level until Call of Pripyat, it’s not multithreaded and can’t be back-hacked to become so, etc etc etc.
In a word, it’s a perfect, textbook example of what happens when a developer tries to create its own engine without fully comprehending what’s involved in the process.
GSC Game World has an ongoing battle with 4A, accusing them of stealing X-Ray source code for the engine that runs Metro 2033. The two do look very similar, but 4A is composed of GSC ex-staffers, including the original X-Ray lead engineer, so it may just be their style.
I agree with Armand in principle – that many engine options would be better than few – but the fact is, making an engine is nothing like making a game. You’ve got level designers and so forth sitting on their hands while the programmers toil on a codebase that won’t be able to actually display anything for months or years. The investment in code and debugging is unbelievable. And the engine is only part of the back-end technical equation. You still have to integrate physics, AI, scripting, pipeline, game logic, and a million other things… things that are often included as part of the package in a middleware engine.
GSC has some really, really good programmers. The AI in STALKER proves that. But if they’d license an engine, they could put those programmers to work modifying it to their needs and coding an even better AI (as well as achieving the no-load open world they originally planned). The years spent on engine design could be spent on polish. They’ve made enough money on the STALKER games that they can afford to take this risk, but I don’t think it’s the right move for them.
As for id, the decision not to license Tech 5 doesn’t surprise me. Now that id is owned by ZeniMax, they don’t want to compete with themselves. There’s also the fact that id’s engine licensing has gone downhill rapidly. Only a handful of games used DOOM3 technology, particularly when compared to Unreal. Tech 5 inexplicably does not support dynamic lighting, either, which is likely to be a thorn in its side moving forward. I think id has ceded the shooter engine market to Unreal.
Meanwhile Valve is likely working on a massive update to Source. That engine has always been stellar but it’s showing its age. They have to implement a DirectX 10 (or 11) renderer and some other features if they want to stay competitive with what’s coming out of eastern Europe and Germany. Oddly, though Valve does license Source, we haven’t seen many games using it. I’ve never been told that it’s hard to work with or anything, so I’m a bit surprised by that.
A 3D modeller who won a weapon design competition with Valve, did an interview recently saying the engine was an absolute nightmare.
I recall him saying it was like sitting on a chair with a spike straight through the middle of the seat… I’ll try to dig it out.
Ouch!
Here we are:
“I see the source engine as a beautiful ornate chair with a spike carved into the seat; It looks great, it’s solid and it will last for a long time, it’s just a pain in the ass to use.”
(In this instalment of How It’s Done, I have interviewed Larolaro, the creator of the Homewrecker and a winner of the famed Polycount pack contest, with his Tank Buster pack for the Soldier.)
The Valve people must be intimately familiar with it, making it easier for them. Vampire: Bloodlines was on Source and it looked (and performed) terribly, so in a way it makes sense that the engine’s not friendly. All those commentary remarks in the Episodes suggest that the thing is easy to work with, but of course, those are the people who made it talking.
This is why I love this site! I learn so much! Did Zeno Clash use the Valve engine or was that something else?
I was thinking about all this a bit last night, I guess considering the cost and time taken to make a game engine, it probably has enough people making them.
You don’t want too much competition for such big ticket objects. Imagine 10 different companies all making giant airplanes and fighting for contracts. May not be a great comparison, but with a small buyer’s market, I suppose you can over saturate it. BTW, I sucked at econ. ( :
Yes! I’d forgotten that one Armand – Zeno Clash, Sin Episodes, Vampire Bloodlines, Dark Messiah of Might & Magic… that’s all that leaps to mind, aside from Valve’s own games. Wikipedia has some interesting info about Source that I didn’t know, particularly that all code items must be written in C++. Who’d have thunk?
The giant airplane analogy isn’t so bad. Still, it’d be okay withe me if we had ten huge middleware engine companies – it’s when small, local carriers like JetBlue decide to build their own airplanes that we should worry.