Kerbal Space Engine - how hard would it be?

[MircoMars]

Hi there,

here's a topic that I fought off thought of for a long time:

How hard would it be to program a custom game engine for Kerbal Space Program?

Could this be an option for 2.0? What would improve, what wouldn't? How many (wo)manhours would this approximately take? Is it even possible to build a proper custom engine in a reasonable timeframe today or are all game developers dependent on having this outsourced to specialized teams like Unity, Unreal etc?

Disclaimer: this question is meant as a discussion starter, totally hypothetical and does neither represent a demand towards SQUAD to port KSP to a custom game engine nor state that the Unity engine underperforms and needs to be replaced.

[kBob]

Well I'm sure it's quite possible but then a new engine is going to have to be  thoroughly debugged and will probably introduce a whole new set of bugs.  A game engine is a pretty major undertaking and you'd have to hire whole new team for it (well unless you wanted all other progress to stop) so the cost issue is going to be pretty high when something already exists; if it was worth doing I'm sure they would have since at least then you can fix bugs yourself and do space physics the way you want, but dang the cost/benefit analysis probably sways towards the keep using Unity side.  Or put it another way: why reinvent the wheel (ok sorry for that problem reference/pun)?

[Vaporized Steel]

Anything can be hypothetical, so in terms of "Can it be done", "Yes we can"

But in practicaility, this will never happen.
Programming a game versus creating your own game engine are two different ballparks.
I don't think the Squad staff is capable or eligible to create their own game engine.
Neither are you and most of the hardcore modders around. You need money, a seperate crew each qualifying with specific IT certificates to finance and produce a custom game engine. I'm pretty sure a larger market is required then the $40,00 of income per purchase for ksp with the additional products in the Squad store.

Game engines are the bread and butter for the developers that program it. So if you want this to be done this new game engine should cover interest for other game developers to work with. Because of this a newly developed game engine is almost never exclusive to just one single game. Except for very large gaming companies and even then a new game engine is almost always reused in other titles under the same brand.

Edited June 14, 2016 by Vaporized Steel

[icedown]

If you want to look at a space simulation with a custom engine, Orbiter is a good example.  The community even developed the directx implementations of the graphics.  While ships can dock together, building custom ships requires extensive knowledge of modeling and programming.  It is a good simulation but the limitations of it are apparent very quickly.  This is the problem with developing your own engine.   I seriously doubt that the ability to put ships together from component parts will be a possibility for it any time in the near future.

[StahnAileron]

An dedicated game engine is a massive undertaking these days. Most game engines are created by relatively big companies (some of which essentially do only the engine and support, no actual games, like Unity). The return on investment is quite low unless you can license it out or get more than one or two games out of it. KSP is a pretty specialized game, so I don't think they'd be able to get mileage out of the engine unless it was HUGELY successful (more so than now). KSP is $30 USD retail (cheaper with discounts). If they had created their own engine, it would've taken FAR longer to release and cost a lot more. Recall that SQUAD started from essentially nothing other than a Unity dev kit. The associated costs would be passed on to the players. Imagine if KSP costs as much a typical AAA title from the likes of EA or Activision? Would you have jumped into KSP if it was $50-$60? (And probably not discounted much, if at all.)

Dedicated engine take a dedicated team to develop and maintain today. The likes of EA, Activision, id Software, Bethesda, and Epic have the resources and incentive to do that (they develop their own in-house engines for their own myriad of projects and for licensing). If the game is buggy and is traced back to the game engine itself being the point of failure, you have to fix the foundation before you can build the house, so to speak. SQUAD doesn't have that kind of resources. Current engine bugs in KSP today are left to the Unity people to fix while the KSP team works on what they can in the actual game.

So yes, it's possible. Hell, I'd argue that it would be in the better interest of KSP due to its ambitions. But is it practical? Not with the currently available resources. There's incentive, but not enough to warrant or justify it at this point. If KSP and things like Orbiter kick-started a new genre of space simulator (and by simulator, I mean real space sim, not "flight sim, but in SPACE!"), perhaps there would be true incentive to develop a dedicated engine.

On the flip side, maybe they could con one of the big engine devs to include an orbital physics package. Something like the current Unreal engine is supposedly flexible for development, but more involved as far as I know. (Then again, I only know of big-name studios that use it.) I wouldn't mind it if someone could port KSP to another engine like Unreal if said engine could support it.

[Spaceception]

When I saw the title, I thought you were talking about merging Space Engine, and KSP together

[mythic_fci]

Switching engines to something better/more flexible than Unity? It'd take lots of hard work, but it'd be doable. Creating a whole NEW engine, on the other hand, is a whole different league of difficulty. Engine creation is something pretty much only major game studios such as Rockstar or Ubisoft do, and even then it's a long and arduous process that requires lots of people, skills, and hard work to pull off successfully. The difficulty level for such a task increases tenfold for a small game developer like Squad, with only a dozen or so programmers.

TL;DR: Unless Squad somehow grows to become as big as Rockstar, it isn't happening.

[Tw1]

1 hour ago, Spaceception said:

When I saw the title, I thought you were talking about merging Space Engine, and KSP together

Yep. Imagine that though... 

[magnemoe]

23 minutes ago, FCISuperGuy said:

Switching engines to something better/more flexible than Unity? It'd take lots of hard work, but it'd be doable. Creating a whole NEW engine, on the other hand, is a whole different league of difficulty. Engine creation is something pretty much only major game studios such as Rockstar or Ubisoft do, and even then it's a long and arduous process that requires lots of people, skills, and hard work to pull off successfully. The difficulty level for such a task increases tenfold for a small game developer like Squad, with only a dozen or so programmers.

TL;DR: Unless Squad somehow grows to become as big as Rockstar, it isn't happening.

That makes this worse is that most engines are old and upgraded over long time and multiple games. 
Unreal is 18 years old. 

[cantab]

I feel a much more suitable game engine is a must to do justice for a "KSP 2" if that's ever made. And I think it should be, because Unity is holding KSP back. (And because they can sell it to all KSP 1's fans!)

If it's an 'off-the-shelf' engine, it needs to be something designed for interplanetary scales. Space is "in" in gaming generally at the moment so prospects might be good, but I don't know of anything specific. Space Engine is probably the closest thing, it's awesome on the graphics and star systems, but I don't think it has anything for vessel physics.

If it's a custom engine, well Squad need to step up their game. Be willing to invest the money on hiring more experienced talent. Not just rank-and-file coders, they need good management too.

My concern is that they risk being late to the party. Like I said space games are big at the moment, but by the time Squad get a KSP 2 out the market could well have moved on as it were.

[DChurchill]

1 hour ago, cantab said:

I feel a much more suitable game engine is a must to do justice for a "KSP 2" if that's ever made.

Like?

[qromodynmc]

Well, Cd  Projekt Red made their own engine after Witcher 1, they were pretty small team too and i suppose they sold less games than KSP.

Still I dont know if it's worth to make another engine for KSP, Overall target audience is kind of small.

[cantab]

43 minutes ago, DChurchill said:

Like?

Like I went on to explain, "something designed for interplanetary scales". So not Unity, and from what I've read about it definitely not Unreal Engine.

[CrashTestDanny]

My managers always ask the "can we" questions  (How people with this level of communication skills got to be managers and stay managers is beyond me, apparently).  This is one of the very few questions that I think of as "dumb" in my world, because the answer is almost always the same; "YES, we can."  Will we?  That's a whole other deal, and hopefully closely tied to the much better question "Should we?"

I don't think we should view Unity as 'holding KSP back' either.  Where would KSP be without Unity?  Is there another engine that was available to the devs when they started that would have enabled them to create this game?  While this question may be of interest from a historical perspective, I am fairly certain that KSP would not be where it is today without Unity.  Unity has enabled Squad to focus on game development instead of physics, or scene management, or probably a hundred other things that Unity handles (like ship editors maybe).  As for switching to another engine, that's really an internal Squad matter as far as I'm concerned, with one caveat; switching to another game engine will require all mods to be re-done for the new engine, so now that the game is established on Unity, moving to another engine would be a HUGE undertaking and would impact a lot of people - that has to be considered.

Danny

[cantab]

Even if Unity was the best choice when KSP started out, that doesn't automatically mean it's the best choice for continuing its development and certainly doesn't mean it's the best choice for a hypothetical sequel.

[Green Baron]

If you want to take a look at game engine source code:

http://www.godotengine.org/

That's a high end game engine, MIT-licensed.

Or:

http://pioneerspacesim.net/

the latter a neat little open source space game.

 

Cheers

[MircoMars]

a quick google search gave me this: http://venturebeat.com/2013/11/05/moonshot-how-kerbal-space-programs-creator-launched-an-indie-darling-from-out-of-nowhere/

HarvesteR pitched KSP as a simple 2D build your own rocket game and then it all took off. First they had a 3 week schedule, with a playable game every 21 days. this was only possible with a pre-built engine and unity was both easy access and free of charge (home version, if you don't make lots of profits with the results). Squad was no software development company and I guess the starting staff had a hard enough time to wrap their heads around 3D modeling, C# scripts and frickin rocket science. when they had a new idea or just wanted to try something out, unity let them do that with relative ease. so, nobody should blame them for starting with unity and then sticking with it!

but KSP has grown, a lot. the players are both ambitious and demanding. constantly bringing KSP, it's framework and their hardware to the limit. Unity has become one of the weak links, as plenty of workarounds integrated by Squad, the way too long 32-bit area and the post U5 update bugfest proved. so my feeling is, if they'd start building KSP as it is now, they would look hard for a better engine than unity.

What I read out of your answers what I agree with:

1. don't change the horse mid race
2. building an engine from scratch is really, really hard and takes a lot of time and skilled programmers, duh
3. building an engine for only one game is not worth the effort
4. the API for mods and assets in general should be changed as little as possible

But this is more answering should they do it and if yes, how...

Edited June 14, 2016 by MircoMars

[JoeNapalm]

 

I'm just gonna point out that "How hard could it be?" has been a question preceding some of the greatest examples of epic failure in the history of ignorance.

I've come to realize that it is human nature to assume anything people don't understand is easy.

-Jn-

[wumpus]

2 hours ago, JoeNapalm said:

 

I'm just gonna point out that "How hard could it be?" has been a question preceding some of the greatest examples of epic failure in the history of ignorance.

I've come to realize that it is human nature to assume anything people don't understand is easy.

-Jn-

Judging from the various abandoned projects lying around: building a game engine isn't all that hard.  Building a game engine worth using is a *lot* harder.  And being able to tell them apart takes a rather long time, so game designers don't want to be the first ones to build a game on the engine.  Consequence: game engine designers need to build games on their own engines (and then *still* have trouble selling them).

Note: the engine from crytek was on absolute fire sale a few months ago.  Presumably it has outrageous performance, outrageous graphical beauty, and outrageous amounts of bugs.  I doubt it was easy to make, but I would think that just the games the parent company made proves it had value.

This isn't to say it isn't that hard.  To build something that KSP can move to you need to clone Unity 4/5, a system that has grown to meet the needs of a *ton* of games and game designers.  You won't pull it off for just one game.  If you still feel the need, find out what orbiter does (presumably something custom, non-made-for-reuse code) and build your "engine" around that.  Orbiter seems like a reasonable target, KSP does not.