Star Citizen - Chris Roberts, lots of spaceship porn, lots of promises

What gets me is that neither of those engines is suited for what they’re trying to do. So what I’ve always wondered was what reasoning they used when choosing it. What was it about that engine that made Chris Roberts sit up and say, “CryEngine is the best engine for this game!” I’m sure he must have explained it at some point, but I sure missed it.

I think they’d have made much more progress by now if they’d just let some of their programmers develop a custom engine. Of course, that’s easy for me to say, all these years later, hindsight being 20/20.

Crytek did the videos and stuff for the Kickstarter in return for the project using their engine.

An achievement in modern games parlance is something like “looked at the achievement list” or “started a game”. I think they probably qualify for one of those.

It’s quite simple. Roberts has always wanted the best-looking games, going way back to Wing Commander and Strike Commander. CryEngine was one of the best looking engines at the time. There was minimal consideration for anything else.

There’s no 7 year anniversary of anything. No 1.0 yet. Jack shit has been accomplished.

Yes, after battle everyone’s general. Back in 2011, what engines were available? Unreal 3 was unusable for large streaming open world with dynamic ToD, Unreal 4 was years away, Unigine was (and still kinda is) unproven (no games shipped), Unity was unusable for this kind of game.
Warhorse were faced with the same decision:

http://web.archive.org/web/20180225010250/https://warhorsestudios.cz/index.php?page=blog&entry=blog_009

And also chose Cryengine, despite it not being designed for RPG. There was nothing better available.

Building custom engine is a possibility, but I do not blame these companies for not wanting to reinvent the wheel (particularly Warhorse guys were fed up with problems they had with custom tech on Mafia 2 development - that game took 8 years to make).

Yeah, engine choices and tech stuff can be a matter of choosing the best of the available options rather than the one true ring, for sure.

A game like Mafia 2, though, doesn’t have to live up to unrealistic expectations like Star Citizen, nor were Warhorse claiming to be the second coming of Game Jesus.

That’s because you’re not thinking like a ptw game developer whose goal is to get whales to pay enormous amounts of money for digital assets of no real value.

Their achievements in that field are truly revolutionary.

I blame them. Reinventing the wheel is painful, but trying to run your boat on wheels is even more painful.

Elite Dangerous which happened around the same time used a custom engine and was fine because they can retain talented developers and a culture of shipping software.

Are you suggesting that CIG should have created an entirely new game engine, from the ground up?

COBRA has existed as a game engine since, literally, the 1980’s. It’s evolved over the decades. They didn’t create a new one for Elite Dangerous.

No, because that would involve CIG.

Either creating a new engine or very substantially reworking an existing engine was a necessary step for Star Citizen to achieve its goals (there are pros and cons to each approach). However, it was also one of the many things they were too incompetent to achieve. So they just put the car in the water and hoped it would float for long enough for them to sell JPEGs.

Right, but they had the access to the code and also the skills required to do the very substantial rework needed to support Dangerous’ use case.

But they did in fact do extensive reworking of the cryengine.

They certainly attempted to do so.

This. I also think CR was more concerned about choosing something that would make for awesome trailers and what not (to sell the idea of the game, and well, get money for development) than something that would be more suited to actually building the game he was promising to build. But I think he thought he could turn the first into the second, and that’s where they hit a wall.

I will ask you too. Which engine should they have chosen if they did not want to build new one from scratch, back in 2011?

So all the stuff in the DF video that the game allows and does that Crysis didn’t is not real?

None. They should have built one from scratch (well, “scratch”, since they could use a lot of middleware to cut a lot of hard work in specific areas like most “custom” engines do), because no engine would do what they were setting to do. That would be the sensible choice if your first concern is building what you promised.

The real question here is “if, to build the game they promised, they needed a custom built engine, why did they choose CryEngine?”

And the answer is simple. It looked good. Great, even. It would make for great trailers and prototypes so they could get money. It was a solid business decision - a marketing decision, even - not a sound engineering decision. And that’s why they’re having engineering problems now, but the trailers are still as good as ever.

That focus - on “selling” instead of “building” - also explains why they’re always adding things to be sold without ever explaining how they tie together in any meaningful way. They talk about “meaningful gameplay” all the time, but the proposed ideas are not only loosely defined at best, but are often clashing with each other in terms of game design.

You think that this would have been less work than reworking the Cryengine?

Less work, I don’t know. Maybe, maybe not. But it would have a much better chance of actually doing what they need/want than what they got now.

I’m not sure I follow your logic here.

On some level, anything that they wanted to write from scratch… they can write from scratch, while still using the CryEngine. Hell, there are big parts where they’ve essentially done exactly this.

While I can imagine a number of cases where integrating with an architecture may impose design restrictions, that could potentially limit your ability to develop a solution if you were unwilling to abandon large sections of the existing architecture, given that they’ve actually done this, it doesn’t seem that they’re limiting themselves in this way. So I’m not seeing how their choice would actually limit their development in the way you seem to be suggesting.

At that point, you’re left with a question of, “Well, if you have to re-write most of the engine, then there’s not a lot of point in paying for that engine,” but they’re still certainly using a ton of the CryEngine in their system. And by using the CryEngine, they were able to leverage CryTech’s existing engineering team (many of whom they seemingly poached from CryTech, if I recall).

It seems like the suggestion of “they should have written the entire engine from scratch” seems to be in direct conflict with the suggestion that the development is taking too long, as I doubt that they would have been able to construct an entire game engine, from scratch, that solves all the problems that they’ve had to solve in their re-working on the CryEngine, while also requiring the additional work of replicating all of the stuff that the CryEngine does that they’re still using.

Even without creating their engine it was still stupid to use CryEngine over UE3. By 2011 a TON of developers not only had experience in UE3 but UE2 and UE1. Compare that to Cryengine and it’s miniscule.

That’s important because if you are trying to do something “that’s never been done before” then you want to be able to hire the smartest engineers who have experience pushing your engine to its boundaries. So at the outset they limited their pool of talent they could pull from. Not only that but if you are an engine developer (in an industry that has stopped proprietary engine development) then being a UE dev is much more marketable than a Cryengine dev.

Let me use a metaphor them. Imagine you’re trying to build a building unlike anything anyone has ever done. Would you build it from the ground up, or would you start on an existing building and them try to change it bit by bit (or build upon its less-than-optimal bases) while keeping the whole thing stable enough that it doesn’t crumble to the floor?

Or perhaps you want to build a plane. You have this really nice and aerodynamic car that looks great. You get it, and you start adding wings, changing the engine, but you must do all that with it running because you have to get from one place to the other.

That’s kind of what’s going on here. CryEngine is not an unopinionated engine - it’s built over certain conventions and assumptions that may work really well for a certain kind of game (FPS, RPGs, action games) but not nearly as well for others (sims, massive open world games, MMOs). The Star Citizen devs are not simply adding or replacing things - they have to fight the CryEngine way of doing things more often than not, some things they can’t remove or throw away because it’s too tightly coupled to other things, and there are bugs and issues they have a hard time finding because the CryEngine is a massive software project that CIG didn’t develop and for which documentation is probably not enough, particularly when discussing its “innards”.

I hope all that clarified it a little bit?