24 replies to this topic

[Mordric]

how do game developers decide on which graphics engines they will use?

[Kira Onime]

eenee meenee miney moe

[Eyepop]

There are various different things that go into it, but the decision can be entirely out of developers' hands, sometimes. Things that go into it:

1) Will going with a certain engine speed up development time?
2) Will marketing or corporate make money from using an engine?

If either is yes, then:
3) Which is best for the above purposes?
4) Which has the required features? (Yes, this does sometimes come afterwards, when dealing with (2) above)

Even if (1) is true, sometimes spending time and money on a proprietary engine is still a good idea, if:
5) Does the engine do too much for you?
6) Does the engine not run fast enough to support certain hardware?

Let me explain (5). In a game like Civilization 5, we can make a lot of assumptions about the rendering that we need to do, so a proprietary engine makes a lot of sense. It doesn't make sense to go with Unity or CryEngine or Unreal, because those can't make the same assumptions that we can; they have to solve a more general rendering problem, so they don't solve our specific one as quickly as a proprietary solution can.

The same concept is why console games can be more optimized (can make assumptions about the hardware), and why DX12, Vulkan, and Mantle are such a big thing (can use specific knowledge to better optimize, instead of using general-purpose paths and forcing the driver to figure it out without all of the knowledge that you have).

Sources: Myself, as a graphics programmer at Firaxis, where we use Unreal (XCOM) and a proprietary engine (Civilization series).

[Whatzituyah]

In my honest opinion its either Unreal Engine that would do mechwarrior online justice or the Havok Engine "Which I think was dubbed hedgehog engine at first hence the real life hedgehog symbol" both were meant for speed I guess Havok wouldn't be a good choice because I don't think it has networking. Unreal Engine however had networking how else was it to perform I.P. to I.P. mutiplayer on the first Unreal? It just improved from there aswell. But we got the crytek engine why got to deal with what we have if they can code they probably can edit the engine to make a totally new engine to fit this games purpose but thats not the case.

[Kiiyor]

I've a few friends in game development, and one thing they all agree on is that you don't just pick based on what you think is best for the game you are developing; you pick what's easiest for you to develop with. Shiny new engines are next to useless if you aren't able to write decent code in them.

Other stuff to consider:

• performance
  
  • Which is wierd - because it depends less on what the engine is capable of, and more on how well you can write code to make it do what you want to
• documentation
  
• maintenance
  
• support
  
• cost
  
• overall features
  
• the programming language it uses
  
• pedigree
  
• ease of use
  
• assets (some engines have asset stores)

And yes, heh, I copied most of this from an old games lecture slide from my universidy days

The long and short of it is, that you pick the engine you are most capable with.

[Destructicus]

"Let's use the engine they made Crysis with!
It'll give the illusion of a AAA experience then we make a game we should have made on Unreal!"

Joking aside, I think Cryengine is a terrible engine to make a F2P game on, it's not very friendly towards lower end machines and it seems to be one of the biggest barriers for PGI in regards to coding things like hit detection and MASC.
I really wish they would have made this with Unreal, of course the shortsighted will scream "LOL BUT THEN IT'D JUST BE HAWKEN"
But really a 3 second google search of Armored Warfare would disprove that.

That being said
My dream engine for MWO would be Frostbite
It would allow for great visuals with larger scale mech combat.
A geek can dream

Edited by Destructicus, 20 June 2015 - 08:31 PM.

[Narcissistic Martyr]

Destructicus, on 20 June 2015 - 08:29 PM, said:

"Let's use the engine they made Crysis with!
It'll give the illusion of a AAA experience then we make a game we should have made on Unreal!"

Joking aside, I think Cryengine is a terrible engine to make a F2P game on, it's not very friendly towards lower end machines and it seems to be one of the biggest barriers for PGI in regards to coding things like hit detection and MASC.
I really wish they would have made this with Unreal, of course the shortsighted will scream "LOL BUT THEN IT'D JUST BE HAWKEN"
But really a 3 second google search of Armored Warfare would disprove that.

That being said
My dream engine for MWO would be Frostbite
It would allow for great visuals with larger scale mech combat.
A geek can dream

PGI used Cryengine because Crytek was trying to make their engine more popular with developers and didn't charge royalties.

[EgoSlayer]

On top of all of the above, there is also the licensing costs of the game engine that often is a determining factor. Some engines require huge upfront costs that a smaller studio can't afford.

[Y E O N N E]

Whatzituyah, on 20 June 2015 - 08:14 PM, said:

In my honest opinion its either Unreal Engine that would do mechwarrior online justice or the Havok Engine "Which I think was dubbed hedgehog engine at first hence the real life hedgehog symbol" both were meant for speed I guess Havok wouldn't be a good choice because I don't think it has networking. Unreal Engine however had networking how else was it to perform I.P. to I.P. mutiplayer on the first Unreal? It just improved from there aswell. But we got the crytek engine why got to deal with what we have if they can code they probably can edit the engine to make a totally new engine to fit this games purpose but thats not the case.

Havok is not a game engine, it's a physics engine. It's a piece of middleware used by developers to simulate physics in games. It supplements the game engine. Source uses Havok for physics and I think most Unreal Engine games also employ Havok. Even the Halo games use Havok for their physics.

[Rushmoar]

Mordric, on 20 June 2015 - 07:35 PM, said:

how do game developers decide on which graphics engines they will use?

Price is the 1st factor that a Dev considers. Then the performance of the engine that is required to make the game that they envision.

Something like this to the best of my knowledge.

[bad arcade kitty]

hey jim, have you ever dealt with cry engine?
yeah, once, in the college
so, it's decided, we pick cry engine

[Peter2k]

CryEngine was free to license back then, and I seem to remember that CryTech threw in support on coding too back then

seems powerful enough, if you can work its magic (and there is the problem)
although the games that came out with it are not that plentiful, especially the more Multiplayer focused ones (Star Citizen looks good, but I have not seen anything that wants me to spent time over there, already am supporting one game that is in development still )



from all the videos I've seen from the Unreal Engine I gathered that Epic wants to make it easy for game devs
I've seen some of their map editors and couldn't stop thinking how fast PGI could make maps that look good by simple drag n drop and some mouse clicks (I know it's a bit simplified, but I do think that CryEngine isn't that easy to handle)

and it seems to scale well fps wise on older systems

[anonymous161]

When pgi chose the current engine used I knew it was just gonna be a down hill roll from their.

[Peter2k]

EgoSlayer, on 20 June 2015 - 10:05 PM, said:

On top of all of the above, there is also the licensing costs of the game engine that often is a determining factor. Some engines require huge upfront costs that a smaller studio can't afford.

sigh too bad Epic didn't do the deal (back when MWO looking for an engine) they do now

Engine is for free
5% royalty

I'm not an expert, but I bet there is a good reason why you see so many games based on Unreal Engine

[Destructicus]

Narcissistic Martyr, on 20 June 2015 - 10:03 PM, said:

PGI used Cryengine because Crytek was trying to make their engine more popular with developers and didn't charge royalties.

PGI used cryengine because "**** it, it's cheap"

[LordNothing]

im kind of disappointed that the option of 'use an in house engine' has been all but axed from developer's tool sets in favor of off the shelf game engines. it kind of gives you somewhat limitless power as far as development goes. you dont have to work around engine limitations as much, as you can design the whole thing with your requirements in mind. it also tended to make games more distinct from each other, such as if you wanted to break out into a new unexplored type of game play. of course it did require the right kind of talent and that talent aint cheap.

[Eyepop]

LordNothing, on 21 June 2015 - 03:50 AM, said:

of course it did require the right kind of talent and that talent aint cheap.

This is exactly why it's not done more. We wouldn't even have proprietary for Civilization, even though it makes so much sense, if we didn't know that we'd be able to ship a couple games and a ton of expansions on the same engine.

Most proprietary engines just aren't worth the time and effort, when one could do the same thing with an off-the-shelf engine.

Concerning CryEngine versus Unreal, it's not so much the engine as it is the developer, in my opinion. Beautiful, optimized games have been made on both engines; sure, fewer people use CryEngine, so there are fewer good examples.

But then, terrible, piece-of-garbage games have been made on both. On the Unreal side, just go take a look at APB Reloaded. They have all of the same problems that people are complaining about here (granted, it is an old, crusty, heavily-modified Unreal engine):
Terrible hit detection
Huge game mechanic iteration time
Years and years between map updates
Really poor optimization

The engine does have some amount of say in the development and end result for a game, but it's not the magic bullet that we seem to be thinking it could have been. Going with Unreal would not have made this game any more what people want than going with CryEngine has.

[Mudhutwarrior]

Question?

Is anybody out there working on an open source engine? Figured you guys would know if there was.

[Y E O N N E]

This game is not mass multiplayer; it's not handling more than 100 players in a single match. You people need to quit calling it that because huge numbers of players in a single match is what "mass multiplayer" refers to, not lots of people having accounts with the game. It's not a case of "different definitions, man," because anybody using the term "mass" for their game when their game doesn't also have that number range of players interacting together is wrong.

And, as far as I am aware, the networking solution used by PGI is not inherited from the engine, anyway. It's a custom job. The optimization problems are also on the developer end. For all we know, PGI locked tessellation on and so we've got water running under entire levels with polygons being rendered that we never see. That was a problem in Crysis 2.

[Eyepop]

Mudhutwarrior, on 21 June 2015 - 07:00 AM, said:

Question?

Is anybody out there working on an open source engine? Figured you guys would know if there was.

Yes.