Experienced Points: EA Crippled BioWare With Frostbite

By Shamus Posted Wednesday May 1, 2019

Filed under: Column 83 comments

In my article this week, I talk about how the decision to use Frostbite has done tons of damage to BioWare. Also, I spent a little time explaining why seemingly-easy features might be hard to implement in Frostbite.

Annoyingly, I didn’t cover all my bases and a bunch of people tried to “gotcha” me with the fact that former BioWare general manager Aaryn Flynn claims it was BioWare’s decision to use Frostbite. I remember hearing about this ages ago, but I dismissed it as obvious corporate ass-covering. Still, if I’d been on the ball I would have demolished that argument before I started in on Frostbite. Maybe I’ll append something to the article, maybe I’ll handle it in an aside in my next column, but in the meantime here’s the problem with Flynn’s statement:


It is rather amazing that so many EA subsidiaries all decided to switch to Frostbite at the same time. In the linked article Flynn says:

[…] we knew that our Eclipse engine, that we shipped DA2 on, wasn’t going to cut it for the future iterations of Dragon Age. It couldn’t do open world, the renderer wasn’t strong enough, those were the two big ones. We thought about multiplayer as well, as Eclipse was single-player only.

To take these points in order:

Okay, Unreal Engine didn’t do open world. Fine. The problem is that neither did Frostbite. When given the choice between a familiar engine that doesn’t do what you need, and an unfamiliar engine that doesn’t also do what you need, why would you choose the latter? “Oh, we needed a garage, so  rather than add a garage to the house we bulldozed the entire property and built a new house that also didn’t have a garage.”

The claim that the FREAKING UNREAL ENGINE isn’t “strong enough” is laughable. Maybe their forked version was getting a bit long in the tooth, but in that case the obvious solution would be to migrate to the latest version of the Unreal Engine and re-implement whatever features they’d added. You’ve got to re-implement features either way, but with Frostbite you have to add even more features, plus break the entire toolchain, force your entire staff to be re-trained on a new engine that was being developed on the fly, and allegedly throw away some portion of your art library because it didn’t work with the new engineI can’t find a source for this, but I read it somewhere. If anyone has a link, please let me know..

The Unreal Engine has supported multiplayer since 1998. It’s got all the multiplayer you could possibly need. I actually tried multiplayer Dragon Age: Inquisition at launch, and it was ridiculously broken. If we’re to believe Flynn, then BioWare ditched a multiplayer-ready engine to embrace multiplayer in a new engine, when multiplayer was a tacked-on afterthought that barely worked.

Even if the Inquisition Team wanted to move to Frostbite, why didn’t anyone in the company backpedal when the migration began causing problems and delays? They shipped three games with this engine, and it never got any easier. Frostbite was always at the top of everyone’s list of grievances when disgruntled employees slipped off and began talking to the press, and these problems have been corroborated by multiple people in multiple studios across multiple projects.

I can’t prove that Frostbite was mandated by EA, but we have to choose one of these:

  1. EA’s executives continued a longstanding pattern of radical shortsighted moves resulting from their lack of domain experience.
  2. Multiple studios making games from sports to RPGs, all decided to embrace the Frostbite engine at once, even though  it was originally designed for linear cinematic shooters and was thus ill-suited to their needs. They continued to use this engine, even when it was clearly against their best interests to do so and was causing delays, bugs, and an overall drop in product quality. NOT ONE of these studios chose to go back to their previous tools, even though they were supposedly free to do so and EA supposedly didn’t care.

I realize the “EA Executives are all dumb” is an attractive narrative and I can respect someone who wants to caution us against embracing the most simplistic and self-gratifying reading of things. Sure, the world is complicated and the EA leadership aren’t actually cartoon villains who want to ruin videogames for the lulz. I’m willing to believe that game developers make lots of dumb mistakes without any help from their parent companyI make lots of mistakes all by myself; I don’t need a billion-dollar behemoth to help me do it..

While I appreciate calls for a nuanced reading of things, in this case I think the easy answer is the one that makes the most sense.

 

Footnotes:

[1] I can’t find a source for this, but I read it somewhere. If anyone has a link, please let me know.

[2] I make lots of mistakes all by myself; I don’t need a billion-dollar behemoth to help me do it.



From The Archives:
 

83 thoughts on “Experienced Points: EA Crippled BioWare With Frostbite

  1. ElementalAlchemist says:

    You are pretty majorly off-base here Shamus. Bioware only used Unreal for ME1-3. Everything else used their own in-house proprietary engines, stretching back to the Infinity engine. Aurora was the step to 3D (for Neverwinter Nights), Odyssey was the console-centric cinematic Aurora derivative for KOTOR and Jade Empire, and Eclipse was more or less a hybrid of the two for Dragon Age (with the more console-focused Lycium derivative being used for DA2).

    1. Shamus says:

      I don’t see how any of this conflicts with what I said. Where am I off-base?

      1. Arne Arnson says:

        1) He was talking about the Eclipse engine, BioWare’s inhouse engine they started with Neverwinter Nights and continued developing (with lots of changes along the way) until Dragon Age 2 (with the exception of the Mass Effects).

        If you look at modding tutorials for these games, you’ll see that the way the engine is structured, they can’t do open world stuff. They’re structured around single mostly self-contained areas hanging together by doors.

        Multiplayer was there, but was mostly ripped out by the time Dragon Age came along. And even then, it assumed the whole group starts together, sticks together and ends the story together. In the case of PWs, you need one persistent server to do the book keeping. The concept of people dropping in and out, ad-hoc multiplayer would not be an easy thing to add.

        Could it be done? Maybe. But it would be a major rewrite of a lot of subsystems, again.

        I for one would have preferred them taking that path, but I have a nostalgic connection to that engine family.

        2) The Mass Effects used Unreal Engine *3*. The current version of the Unreal Engine is 4. And it works very differently.

        Moving from Unreal Engine 3 to Unreal Engine 4 would also make a lot of relearning necessary. It’s not unlike the move to Frostbite.

        1. PPX14 says:

          Hmm it seems that Shamus’ point isn’t diminished by Eclipse et al not being based on Unreal, if ME1-3 were on Unreal so Bioware had plenty of Unreal experience .

          But if your point 2 is correct, then it does look as though it largely invalidates his point. Edit: that specific point, not the one about all of the studios mysteriously co-ordinating their change to Frostbite.

        2. RoboticWater says:

          “Moving from Unreal Engine 3 to Unreal Engine 4 would also make a lot of relearning necessary. It’s not unlike the move to Frostbite.”

          It very much is unlike Frostbite. First of all, the jump from UE3 to UE4 is not nearly as big as you think it is. Having developed in both, pretty much every change in UE4 has only made UE3 features more intuitive, and that’s on top of giving you direct access to the source. Is it a different workflow, yes, but very familiar nonetheless.

          Beyond that, Unreal is practically an industry standard, so there’s tons of support and examples out there from developers working on every type of genre, not just military FPS. There’s simply no way BioWare could have had the same issues implementing a save/load system with the kind of documentation and resources out there. I feel very confident saying that developers would probably have an easier time implementing a Frostbite-like level streaming system into Unreal than implementing all the RPG stuff into Frostbite.

          And the standardization also means that hiring new talent is infinitely easier. No need for long on-boarding processes, teaching engineers all of Frostbite’s quirks; most people know Unreal coming in (even straight out of college).

      2. ElementalAlchemist says:

        My point was the problem is not really that they had a bunch of generalist programmers and had to add stuff to Frostbite that the didn’t need to add to Unreal (which is not exactly true anyway). They had over a decade of experience building their own engines in-house by the time they started work on Inquisition and transitioned to Frostbite. Their real problem was twofold.

        They retrofitted Frostbite on three separate occasions almost entirely independently. The Inquisition team did help out to some degree with Andromeda and Anthem, but it’s pretty clear that the wheel was reinvented for each project. If they’d done it once at the outset and used the same codebase for all three projects like any sane, rational person would propose, I doubt Frostbite issues would be a major topic (beyond the usual client-side performance issues and so forth).

        They appear to be plagued with seriously half-assed project management. It’s especially apparent with Andromeda and Anthem, but Inquisition had many of the same issues, they just managed to salvage a more cohesive game out of it (although not by much) and that masked a lot of the trouble. It’s clear this fed back into the Frostbite development side of things. The problem wasn’t that it was hard as much as they had to scrap and redo stuff over and over because nobody had any sort of plan. It actually says a lot of their skill that they were able to salvage anything remotely playable in the case of Anthem (which was more or less made from scratch in under 18 months).

        So yeah anyway, Frostbite was not the problem. It’s just an easy excuse for bad management.

    2. Guest says:

      Also, Frostbite has solid multiplayer. EA and DICE have been releasing Frostbite multiplayer games for half a decade.

      You are seriously off here Shamus.

      I agree that it’s almost certain EA influence Bioware to use Frostbite. They want all their devs to use it, to promote it, to sell licenses. That almost certainly had a negative effect on the company-including documented issues like Frostbite having no third person camera by default, and being a harsh conversion, because it’s pretty singularly designed for FPS games heavy on spectacle, and also multiplayer, and these are designed to be viewed in the player camera.

      But you really messed up mentioning multiplayer. FB has had a lot of MP support, for a long time, and I’m 90% sure I’m reading you wrong, it’s a really jumbled paragraph. Maybe revise this one? I don’t think I’ll be the only one to misread it, because it’s not like MP was a focus for Andromeda, so I really doubt that was where you were going, and the info there is just, wrong? Are you being sarcastic? What’s with the switch from Open World discussion, where, yeah, you’re right, FB isn’t designed for an open world, to MP, where, you’re dead wrong, FB is designed for multiplayer, and has years of support behind it in that area, as opposed to Eclipse, which has one release with patched on MP to build off. FB has the vast majority of Battlefield in it’s development history, all using multiplayer, nearly all using FB for MP. It’s a non sequitur at best.

      Second point really doesn’t support your argument, and includes factual errors. For once, not trying to bust your balls here, I just think you might want to repolish this one.

  2. Yerushalmi says:

    “Oh, we needed a garage, so rather than add a garage to the house we bulldozed the entire property and built a new house that also didn’t have a garage.”

    From Dave Barry:

    The reason we bought a motorboat is, we needed a new kitchen. Our current kitchen has a lot of problems, such as a built-in Colonial-era microwave that we think might not be totally safe because it can cook food that is sitting as far as 15 feet away. We had spent months striding around our current kitchen, making sweeping gestures and saying things like, “We’ll move the sink over there!”

    What a pair of goobers. As you experienced renovators know, it’s easier to construct a major suspension bridge than to move a residential sink. Thousands of homeowners who embarked on sink-relocation projects during the Eisenhower administration are still washing their dishes in the bathtub. My wife and I kept running into people like this, people with plaster dust in their hair and hollow eyes from spending their wretched nights sleeping in the garage and their bleak days waiting desperately for workmen who inevitably made things WORSE. “We have no telephone or electricity or water,” the Renovation People would say, “and on Monday a man is supposed to come and take all our oxygen.”

    This was discouraging, but we really needed a new kitchen. Finally we said, OK, if we don’t do it now, we’re never going to do it, so we decided to bite the bullet and: buy a motorboat. Our reasoning was, “Hey, if we have a motorboat, we’ll have Family Outings where we can experience Togetherness and possibly crash into a reef and sink, and then it won’t matter about our kitchen.”

  3. Mako says:

    I really don’t see how any statement made by a higher-up from EA/BioWare in the last 7 years or so can be taken at face value. And yet some people still act as if modern BioWare is the same company that it had been in 2003.

    This is a company that had been twisting facts and insulting the intelligence of their audience (in their PR speak, I’d argue that’s also true for their later games to a degree, but that’s a separate topic in itself, and one covered at length on this blog) since at least 2012 (that’s how far back I can trace it, anyway). Of course they are covering their ass. It’s the way they do things now, and they have for a good chunk of the last decade (possibly for more than that).

    1. kdansky says:

      > I really don’t see how any statement made by a higher-up from EA/BioWare in the last 7 years or so can be taken at face value.

      If politics taught us anything, then that a large number of people will accept any boldfaced lie at face-value as long as it supports their pre-existing beliefs.

      It is quite shocking how easily EA or Ubi or Blizzard get away with making shit up and don’t get called on it.

    2. Sleeping Dragon says:

      Shrug, it’s all a matter of context. For example, EA sends out a memo along the lines “over the next X years we want all our studios to switch to Frostbite for all their releases”, or something a bit more vague like “the company strategy going forward is to promote products made with in house tools”, and at some point whoever is in the decision making position at Bioware at the moment decides that it’s time to bite that bullet, because seriously, what else are you going to do? We all know how it works. Every single problem with your development cycle is going to be blamed on not using Frostbite, every single cost overrun is going to be blamed on not using Frostbite, every single time you need to ask EA for something you’re entering the discussion from the loosing position of “those guys who refuse to switch to Frostbite and keep costing us money on Unreal”.

      Especially since, like Shamus keeps pointing out, corporate CEOs don’t understand the nitty gritty of game dev. Yes, you can show them the pretty forest map you made in Unreal, but they’ve also seen the pretty stadium and the pretty bunker and the pretty beach all made in Frostbite, they’re not gonna know the difference, especially since they’ll be told Frostbite can do anything with just a little fiddling, and at the same time they know Unreal is costing them money. Remember, it’s in house made so there are people within the company with strong interest in pushing it, I’m not even saying they’d be lying but especially with things like making or using a game engine there is a world of difference between “I believe it’s possible” and “I can and know how to do it and it’s going to work and not cause issues”,

  4. Gethsemani says:

    I imagine the decision to switch game engine went something like this:
    Bioware: “We want a game engine that can develop good RPGs, with open worlds, multiplayer support and all the stuff we had before. So we were thinking maybe using Unreal, Epic has made us a good offer.”
    EA: “Yeah, sure. Absolutely. You can totally use Unreal, just give Epic a call if that’s what you want. But before you do, let me propose this: You can use your in-house studio Frostbite for free. That’s real free, no strings attached. It gets better too: We’ll give you direct access to the Frostbite teach team and support assets we are setting up RIGHT NOW. They’ll help you modify the engine however you like, no charge.”
    BW: “That sounds good and all…”
    EA: “But? Come on, talk to me, I just want what’s best for you.”
    BW: “We hear that Frostbite doesn’t do anything but Battlefield games well. That it lacks open world capability, no world save states, no dialogue trees or branching options, not even an inventory system. These are really important things to us. It is what we need to make a good game.”
    EA: “I hear you. Loud and clear. So look, Frostbite isn’t perfect, I’ll be honest about that. It needs some work to suit you. But look at FIFA, it runs on Frostbite and it is a great soccer game, right?”
    BW: “… Right.”
    EA: “That’s what we’ll do for you too. Just sign here and we’ll put you in direct touch with the tech team. They’ll work real close with you to make Frostbite the engine of your dreams and you don’t need to pay a dime or do any ad hoc solutions on your own. Not like Epic who’d leave you hanging if you need to change their engine.”
    BW: “Well, it does sound good. And no cost you said?”
    EA: “No cost. Here, have this number to the chief tech guy on Frostbite. Call him, see what he has to say and then decide if you want us or Epic.”
    *BW goes away to make the call and comes back*
    BW: “He sounded real enthusiastic. Like he really knew what he was talking about and that we’d get full support for any changes we need to do.”
    EA: “Didn’t I tell you? So just sign here, here and here. Oh and this here must be signed in blood.”
    BW: “What? Blood?”
    EA: “Just a legal formality. You know, bureaucracy, am I right?”
    BW: “Right” *signs in blood*
    EA: “Great! We’ll get you the best engine you ever used! Glad to be partners!” *shakes BWs hand* “Oh and uh… By the way, this isn’t a problem or anything, but just so you know there are a few other studios lined up to use Frostbite.”
    BW: “What do you mean?”
    EA: “Just that the tech staff is scaled to handle maybe three concurrent projects. Right now they are up to… “*fades to murmuring*
    BW: “To what? I didn’t hear you.”
    EA: “A dozen concurrent projects. But don’t worry, I promise you’ll get priority. You have it on my honor as the Worst Company in America.”
    *Smash cut BW’s horrified face*
    The End

    1. Olivier FAURE says:

      That sounds about right.

  5. Ninety-Three says:

    Multiple studios making games from sports to RPGs, all decided to embrace the Frostbite engine at once

    Your timeline doesn’t make sense. Here is the total list of games/stuidos using Frostbite.

    EA Digital Illusions CE, the cinematic linear shooters people, used it for a dozen games stretching back a decade.
    Need for Speed adopted it for a game released 2011, continued 2013, 2015, 2017.
    Medal of Honor (cinematic linear shooter) adopted it 2012.
    Plants vs Zombies Garden Warfare (the weird shooter spinoff) adopted it for a 2013 release, continued 2015.
    Visceral Games adopted it for a cinematic linear shooter released 2013, continued 2015.
    Bioware adopted it for a game released 2014, continued 2017, 2019.
    A random golf game from the Madden studio adopted it 2015.
    Madden adopted it for a game released 2017, continued 2018.
    FIFA adopted it for a game released 2016, continued 2017, 2018.

    When exactly is this “all at once” you allege is due to executive meddling? And if EA really is mandating the use of their cinematic linear shooter engine, why can we put it on at absolute most two cinematic linear sports studios, PopCap’s weird shooter experiment they had no preexisting engine for, and Bioware? If there was a command from on high to use Frostbite even where it doesn’t belong, EA has a lot more studios than that, we should be seeing it everywhere.

    The answer that makes the most sense to me is that over a period of four years, literally three studios decided to use Frostbite for defensible reasons (cinematic sports studio, cinematic sports studio, guys with no shooter experience making a shooter), and then Bioware took crazy pills.

    1. Ninety-Three says:

      Minor too-late-to-edit: PvZGW was 2014,2016, transcribed that wrong.

    2. Decius says:

      More than one studio per year averaged across 2011-2015.

      1. Ninety-Three says:

        That five year period gets you one racing series, one sports game, one Bioware game, and four shooter games.

        You will understand if I don’t find this compelling evidence for “EA forced all their studios to use Frostbite”.

        1. Mistwraithe says:

          What would be compelling evidence? Presumably all of these studios had other engines they used before switching to Frostbite? Did any EA studios switch away from Frostbite?

          Your list actually seems pretty compelling evidence to me that there is significant pressure within EA for their studios to switch to Frostbite.

    3. Geebs says:

      If you factor in different development cycle lengths it’s pretty darn simultaneous.

      I definitely seem to remember there being talk at the time of this being a deliberate companywide decision to adopt Frostbite but I doubt I could find a reliable link.

      Nevertheless, even if EA didn’t actually mandate the use of Frostbite, they are certainly alleged to have cut Bioware off at the knees by constantly poaching anybody who became competent with the engine to work on FIFA instead. If anything, that’s a far worse case of executive meddling.

      1. Ninety-Three says:

        If you factor in different development cycle lengths it’s pretty darn simultaneous.

        When? Bioware games take longer to develop than yearly sports titles and DA3 already predates the sports games, so those can’t be coming out of the same timeframe. If we assume Need for Speed 2011 got churned out very quickly, you could make a case for the edict being 2010ish, leading to Bioware, NfS, and three companies making shooters for which Frostbite was a totally reasonable call.

        Bioware, a racing series, and three studios with shooters, all adopted a shooter engine. That’s incredibly flimsy evidence of EA meddling.

        1. Geebs says:

          Actually it wasn’t that hard to find EA corporate flat out saying they planned to standardise on Frostbite

          I still think the deliberate brain drain of Frostbite-savvy employees is the more important evidence of EA’s management meddling.

          1. Ninety-Three says:

            The article describes a gradual transition towards more use of Frostbite. It’s dated 2016 (5+ years after DA3 started development) and EA is still making non-Frostbite games in 2019. This is not the “EA made Bioware do Frostbite” I was promised, and it’s certainly not pretty darn simultaneous.

            1. Geebs says:

              I think we’re arguing past each other based on perception of time scale.

              From my point of view, given the huge size of EA, the number of people involved, and the fact that games of the kind that EA publishes are in development for years before they are either announced or released, events happening within a five year span are effectively simultaneous.

              I think you just have a different view of Institutional Time from me.

              1. Jeff says:

                From his username and sense of time, I’m guessing they’re in their mid-twenties and haven’t calibrated their judgment of time to being outside of the education cycles yet.

                4 years is forever to a young adult, that was their entire university experience! High school was forever and a decade! That’s like almost half their life!

                Then ten years later when they’re in their mid-thirties they suddenly realize things that just happened “a little while ago” was actually over ten years ago and where has the time gone?

            2. Daimbert says:

              From the original article and based on my experience in a different development environment, what I suspect happened was this:

              EA leadership discovered that they had a lot of different studios using a lot of different engines to do very similar things and said that they wanted them all to go to one engine at some point, to reduce separate licensing fees and to make moving teams from one product to another easier (as all teams would at least know the engine and the art teams already knew how to generate art assets that worked in that engine, etc, etc). They probably at this point didn’t have any real opinion on which one that would be.

              The internal Frostbite team, for various reasons, wanted it to be their engine and ended up with the internal political clout to do so (the CTO being a booster is a big deal). This probably included them overpromising a bit what it could do, either because they wanted theirs to win out or because they didn’t think that adding the things it didn’t do would be as hard as it was, nor that it would be as hard to use as it seems to be.

              EA then said that every studio should move towards Frostbite as their main engine.

              The Bioware team quoted knew that their existing engine was getting long in the tooth and didn’t do multiplayer well, and so decided that since they likely needed to use a new engine anyway and Frostbite did the multiplayer stuff they wanted to do, it only made sense to do it now instead of moving to a new engine now and then moving to Frostbite later. Thus, it really was their decision.

              Using Frostbite was harder than anyone thought it would be, even for multiplayer, which is why Inquisition’s multiplayer was poor even though the engine is capable of more, as multiplayer was secondary for Inquisition and so more time was spent getting everything else to work and the multiplayer itself didn’t just work with little changes. This then had a huge impact on those games.

              It also seems like code or knowledge sharing didn’t happen, whether that was because of some internal politics about incorporating changes into in basic Frostbite code, internal rivalries, or just a lack of time.

              So it’s likely all true: EA wanted everyone to go to Frostbite, but left it up to the studios to decide precisely when that happened, making the use of it in a specific game indeed their choice.

              1. Sleeping Dragon says:

                Yeah, pretty much what I think happened, this kind of company policy is not instituted overnight but it is being pushed on people. An important thing we should keep in mind is that to understand the process it’s better to think positive rather than attribute malice, I’m fairly certain nobody at EA chuckled evilly at the thought of switching all those devs to a toolset they don’t know and ruining all the games. I mean, when you look at the idea of it there is a lot of benefits of switching to an in house engine, and I don’t mean just “saving money”.

                As some examples off the top of my head: If studios are using different tools it limits how much they can assist each other, they also keep encountering different problems or have to find different solutions to the same problems because they’re using different toolsets. If we have an in house engine we are not limited in how much we can alter it, we have experts who know how to alter it because they know how it it was made, and we also know exactly what it is capable of. If we encounter problems everything is on site and can be kept in company, we don’t have to turn to outsiders (anybody remember PUBG suing Epic?) or even advertise another company on our splash screens.

                Again, the practical application of this clearly ran into some problems.

  6. Cubic says:

    I know a guy who worked at Dice EA (who made Frostbite for their little game). They seemed pretty gung-ho about Frostbite and he gave me the impression that they were going to roll it out all over EA. Seems not entirely unlikely since the Dice founder also was … CTO? for EA at the time. Then my guy changed jobs and the news dried up. Well, take this gossip for what it’s worth.

    “EA’s executives continued a longstanding pattern of radical shortsighted moves resulting from their lack of domain experience.”

    Actually, I would say it was a fairly common strategic move, accepting some short term pain in exchange for better technology in the long term. Unfortunately, they seem to have bungled the transition so far. (It’s usually not easy.)

  7. Ashen says:

    Just to add to what others are saying, Respawn was free to use Source for all of their Titanfall games and is now making that Star Wars game based on Unreal. And those are the kind of games where switching to Frostbite would make the most sense. It really doesn’t seem like EA mandated or even pushed hard anything really. On the other hand I can totally buy into Bioware upper management making a stupid decision.

    That doesn’t absolve EA entirely of course. If the reports are correct, the Frostbite team heavily prioritized implementing features for the games that actually sell, ie. their sports lineup, leaving Bioware in the dust.

    1. Infinitron says:

      Difference: Respawn was a privately owned third party studio until recently, not an internal EA studio.

  8. Mattias42 says:

    I asked this over at Escapist, but think it got lost in the shuffle:

    Regarding the door problem thing… couldn’t you just have added a small box near the top of the door frame, representing one of those door closer things?

    Door opens, about five seconds later, it closes. Meaning that the default state of a door is still closed, but that animation encourages people to go in or out quickly too.

    I mean, it’s kinda like how all doors in games are double hinged. Sure, it doesn’t make much sense that every door suddenly has one of those things, but most players won’t ever consider it and it makes things work just a tiny bit smoother.

    Still, I’d love to hear what the actual solution turned out to be.

    Don’t want to seem pushy, but it occurred to me pretty much on the spot as a ‘cut the cord’ type solution, and I’m deeply curious if Shamus slash you guys think it might have worked or not.

    1. Shamus says:

      “Doors” was a specific example of a general concept. You could attach various behaviors to any object in the world.

      “When clicked on, move 2 meters to the right.”

      “When clicked on, rotate this other object 90 degrees.”

      “When bumped into, vanish entirely.”

      I called them doors in the article, but it was more a system for making lots of different kinds of objects move, change texture, light up, animate, etc. Anyone could set up these behaviors on any object they liked, but the problem was that the server wasn’t equipped to process those scripts and propagate the cascading state changes, so when newcomers arrived the area was always in the default starting state.

      1. Mattias42 says:

        Oh~, fair enough. Thought you meant the, well, doors literally. My bad.

        Thank you so much for the answer!

  9. Infinitron says:

    Is Frostbite really designed for “linear cinematic shooters”? Battlefield isn’t Call of Duty. Its focus is multiplayer in large maps with vehicles and aircraft.

    1. Decius says:

      Those large maps are much more linear than most people think. They certainly aren’t ‘open world’ maps.

      1. Gethsemani says:

        Most Battlefield maps are large when compared to your average FPS map. They are small when compared to just about any open world game map. Frostbite was designed to handle large, mostly static FPS maps with incredible fidelity, not huge open world maps with mediocre fidelity and a need for shifting world states. To my limited understanding they are like night and day in terms of engine requirements.

        1. Sartharina says:

          Battlefield maps have drop-in multiplayer and destructible environments. I don’t know enough about development to know if it’s relevant, but the world state needs to be ‘saved’/synchronized and loaded onto the client of anyone joining the game in the middle of a match. As far as I can tell, Engines are mostly about handling game rendering and physics than the spreadsheets of an RPG.

          1. Shamus says:

            I haven’t been able to keep up with this thread today due to real-life obligations, which is a pity because there’s a lot of programming talk going on and I’d love to be in on that. But as a sort of half-assed response to points people have made at various points:

            Getting back to the stuff I was talking about in my Escapist article, and the kind of thing from the write-up at Kotaku.

            Battlefield’s destructible environments do provide “open world” in the sense of being big, but the problem goes beyond real estate. Can it maintain that destruction between sessions? Based on what the devs have said, this is part of the problem. The engine wanted to load a game from a static map and THEN you can blow stuff up on it. The problem is that in an RPG, things need to change around the map based on previous visits. Doors need to be locked / open, machines need to be fixed / destroyed, NPCs need to be killed, added, moved, or have their appearance change. Can our code gain access to the world state after the world is loaded but BEFORE things start rendering?

            It seems trivial, but like I said in my article, seemingly trivial things can be maddening to pull off if you’re working against the engine.

            Note that this doesn’t mean that the BioWare leadership is blameless[1] or that BioWare’s programmers did a great job[2], just that I don’t want to let the EA execs off the hook for their part in all of this.

            [1.] By all accounts, they were AWOL or asleep at the wheel.

            [2.] Why were Inquisition, Andromeda, and Anthem released in descending order of polish? Even if the engine is garbage, shouldn’t more time = more experience and polish?

            1. DerJungerLudendorff says:

              To your second footnote: since Anthem (and to a lesser extent Andromada) were horrifically mismanaged, didn’t have a clear overall vision and were running out of time before they properly finished the game, this isn’t really a suprise.

              Also, the stress and problems really seemed to be piling up by the time Anthem was actually being made. They were constantly losing more long-time veterans, and most developers were pretty much spent and physically and mentally worn down to the nub.

              1. Cubic says:

                The two postmortems were interesting.

                Count the number of management failures in the Andromeda story. They were casting about for what to do for three years or so. Interesting that they had mock reviewers for guidance before the release but these didn’t work.

                https://kotaku.com/the-story-behind-mass-effect-andromedas-troubled-five-1795886428

                Anthem did a lot of technical dreaming that didn’t pan out and basically had no firm idea of what they were going to do and how. For four years.

                https://kotaku.com/how-biowares-anthem-went-wrong-1833731964

                Seems to me these studios should work out the basic design plus gameplay (and other risk factors) before going to full production. Do a few toll gates on a smaller scale first, like story, modelling, tech and tools, gameplay, perhaps some more. Then the full-scale production project can start with something more straightforward to implement and perhaps be aware of the main risks. (They might need to iterate anyway but it seems like these projects had big problems out of the gate.)

                That way, they can (1) run more prestudies than projects and (2) leave those prestudies in development hell until they graduate or are shut down. I guess it’s basically the Hollywood model.

    2. Gurgl says:

      This is where I really can’t follow the article. Independently of what the Frostbite engine might have been initially intended for, the recent iteration absolutely shines when it comes to rendering huge environments, the idea that its biggest strength is linear shooters is presposterous.

      And I have no clue where this “open-world” obsession comes from. BioWare hasn’t made a single open-world game ever before Anthem, all of their RPGs have been done the traditional way, as in boxed areas linked by transitional checkpoints. In Dragon Age Inquisition and Mass Effect Andromeda, the boxes have become huge, so this completely goes against the idea that the new engine got in the way of a larger game world.

      1. RichardW says:

        The “linear shooter” comment is probably referring to single-player content that’s used Frostbite before. Most DICE campaigns are a bit more stripped down and rely on basic checkpoint systems for saving, instead of the more robust character states and inventories that’d have to be kept track of in a role playing game. The problem for BioWare wasn’t map complexity but what the engine could support saving, would’ve needed to expand that capability a whole lot.

  10. DeadlyDark says:

    In the quote they are talking about Eclipse engine, not the Unreal one, am I wrong? Or your point is that, the problems with Eclipse engine are more easily solvable with the Unreal engine and not the Frostbite?

    1. methermeneus says:

      As I understand it—mostly from comments here and on The Escapist, so don’t take this as definitive or anything— the Eclipse engine is a fork of Unreal 3. The point is that using Eclipse or Unreal 4 would at least provide a familiar toolchain and easier solutions while they solved their problems, not that there wouldn’t be any problems to solve. (Likely fewer problems, though, using a generalized engine with widespread use and therefore lots of documentation and an experienced community of users.) It might not have been a big deal compared to other problems facing Inquisition and Anthem (and probably Andromeda, although I haven’t read as much about that) if there had been some coordination among the teams who had made the conversion (there’s rumors they were actually forbidden from working together), if they’d gotten decent technical support from DICE (they had their hands full helping FIFA port to Frostbite, and EA understandably made that cash cow a priority, but not-so-understandably still expected Bioware to release a timely and polished product in spite of tech support wait times counted in months), or even if they’d just been allowed to keep working with the devs they had instead of having to send their best and brightest off to help the FIFA team.

      1. ElementalAlchemist says:

        Eclipse was an in-house Bioware engine. It’s origins lie in their prior engines, primarily Aurora (Neverwinter Nights), but also with some legacy elements from Infinity (Baldur’s Gate). It has absolutely nothing to do at all with Unreal.

        1. Shamus says:

          This is confusing. Everything I’ve read from BioWare devs has them talking about Unreal Engine, or comparing to Unreal Engine, or complaining how Frostbite is not like Unreal.

          After reading those sorts of comments for years, I just sort of accepted it as a given that they must have been using Unreal before Frostbite. In fact, when I read the linked article and they called it “Eclipse” I figured it was just a fork of Unreal. (Like, they took the conversation / questing engine of NWN era games and dropped it into 2003 era UE, and named the resulting chimera “Eclipse”. Something like that.)

          But yeah. Eclipse is apparently an unrelated species of game engine. What was actually happening is that Mass Effect began life in Unreal and Dragon Age began in Eclipse. Apparently when BioWare spun up a new studio for Mass Effect, the new studio began with modern tools, while the Dragon Age team stuck with what was already familiar.

          This means that Aaryn Flynn was right that they really did need to move to a new engine. They had to choose Unreal or Frostbite. Frostbite was obviously a disaster in the long run (and I still suspect the widespread engine adoption was the result of pressure from corporate and not an accident) but from the standpoint of someone in 2011, the choice wasn’t NEARLY as obvious as I made it out to be above.

          Which means I wound up building a pretty good argument about company-wide engine adoption on top of a TERRIBLE example that undercuts the point I’m making. (I should have just gone back to Andromeda. Yes, it gets old hammering on the same example again and again, but it’s really handy when you don’t want to have to spend half a day and half a column on researching and explaining something as dry as engine lineage.) I might edit this post, or I might just write a follow-up. We’ll see.

          1. ElementalAlchemist says:

            A few corrections.

            DAO actually began life in Aurora (i.e. the Neverwinter Nights engine). The entire game was prototyped in Aurora and was fully playable to completion while they developed Eclipse (they actually showed off some of the Aurora version when they first announced it way back at E3 circa 2003-2004. I distinctly recall pics of what became the battle at Ostagar). As I stated in a reply above, all of Bioware’s in-house engines are derived from the same codebase starting with Infinity, used for Baldur’s Gate 1/2. Aurora was their first full 3D engine for NWN, which was refined into Odyssey for KOTOR and Jade Empire. Eclipse was really the ultimate tool for the style of games they had been building to that point. The cutscene editor in particular blew something like Unreal 3’s Matinee out of the water (even Unreal 4’s Sequencer is arguably inferior in some ways).

            They didn’t start a new studio for Mass Effect (not the OT anyway). ME1-3 were developed at Edmonton, the “real”/original Bioware. The second studio they opened was Austin to develop SWTOR. While that was planned while they were still independent, I don’t think it actually eventuated until 2008, the year after EA bought them. Subsequently the Montreal studio was added. They did some DLC work for ME3 and the multiplayer (having previously internally pitched a spinoff ME FPS). They made the Omega DLC entirely. Then of course they got Andromeda as their first real full size project.

            This is confusing. Everything I’ve read from BioWare devs has them talking about Unreal Engine, or comparing to Unreal Engine, or complaining how Frostbite is not like Unreal.

            What you are seeing there is the Anthem team, which was the former Mass Effect team. Pretty much none of them would have ever seen Eclipse, as that was the province of the DA team. And by the time ME3 rolled around and they were thinking about Anthem and a switch to Frostbite, a number of the old hands may have moved on, and a lot of new hires would have come in that spent their time entirely with ME and Unreal 3.

  11. tremor3258 says:

    I could blame more if this was the first game with the engine, and I put a lot of Andromeda’s issues with their ‘storytelling’ resulting from the engine.

    The after-action review, even internally, should have given Bioware leadership more points where they needed to improve their engine team, if they couldn’t get the help from EA they needed. It doesn’t seem project planning for Anthem adapted or had that as a consideration when they could pencil in ‘Bioware Magic’.

  12. guy says:

    The other theme I noticed in the articles about the three games is that all of them spent a really long time in preproduction with the design in flux, then rushed to release in a short window. Anthem apparently having it particularly bad with a constant back and forth on whether there’d be flight in the game that resulted in repeatedly redesigning levels to account for its presence or absence.

    1. Pax says:

      Yeah, it definitely seems they have a management problem if nothing else. Especially if one of the management decisions was “Let’s use Frostbite.”

  13. Hector says:

    I disagree with the article on one point. Shamus has consistently suggested having experienced design, programming, or art production employees in executive positions. I am not so sure that’s the solution. They certainly shouldn’t be excluded by any means but there’s no inherent reason they’d be outright better.

    The kinds of issues that EA ran into regarding Frostbite were not complex and required no long experience or deep industry knowledge. They just needed to listen to their Dev teams and plan a rollout over several years. That they apparently didn’t is a business culture problem, not a knowledge problem. And unfortunately, I am not sure that, say, Bioware’s management would have done better given their own issues. So I don’t think its just a question of relevant experience.

  14. shoeboxjeddy says:

    I’m a tad confused about the description of Frostbite as for “linear shooters” considering Battlefield is a series that has always run on Frostbite and is about 90% multiplayer by volume. I definitely get how there were problems for adding open worlds and quests and dialogue trees and etc to that engine, but multiplayer would be one of the defining features of that engine…

    1. Gurgl says:

      Admittedly, most of us are not world-class coders or programmers, but quests and dialogues sound like the tamest thing to brute-force into an engine.

      1. Richard says:

        “Dialog” means talking to NPCs

        That means the player gets up-close and personal with up to four or five NPCs, and has time to evaluate exactly how uncanny and weird they look, and how it’s strange that one just said their entire family and almost everyone they knew just got killed by a rampaging Bad Guy yet they’re smiling at me. And why is that one slowly folding themselves up and sliding away?

        If an engine is designed for your interactions to last a few seconds at long range, with the closest you get being looking through a sniper rifle scope…

        1. Sleeping Dragon says:

          Dialogue in this case can also mean choices and I remember hearing that Frostbite was not prepared to handle things like “go to the mob boss and either threathen him or join him, depending on which option you picked you’ll have a different merchant available at your base and you will enter the prison in a different way, also some companions will comment on it with different lines for each option, oh and if it’s the third time you went with the ‘evil’ choice the paladin companion will leave you unless you’ve romanced them”.

  15. Steve C says:

    I don’t believe that in World of Warcraft, it ever was set up to “Instead of attempting to update the game with the position of hundreds of other players, the server just picks the dozen or so people closest to you.” In WoW it hammered all the clients with all the data and let them deal with it. Too bad for you if your connection couldn’t handle it. You got massive lag then disconnected. Which I both witnessed happen to others and experienced myself. (BTW the number is off by an order of magnitude too. Raid sizes were 40. So two raids fighting it out would be 80. And then there’s times when two raids enter a hostile city to kill a faction leader. Which easily pushes it to 80+all the players sitting in the city. Therefore I know WoW could handle 120+ players as I’ve seen that.)

    However Guild Wars 2 absolutely had this “feature” of only displaying X number of characters. It was a serious problem in World PVP. My guild folded for precisely this issue. The developers never fixed it properly. And it could not handle 40 people. It was really bad.

    1. Zagzag says:

      GW2 did fix that problem, but not for quite a while after launch. These days you can configure how many characters should be fully rendered and how many above that threshold should be replaced with generic low-poly models instead of being entirely invisible. They also made nameplates visible on character models that aren’t being rendered, which is the critical part for PvP. It’s no longer possible to have an entire enemy server run you over in WvW with no possible way of seeing them coming.

      1. Steve C says:

        My guild waited until the second fix. The first fix was something I don’t remember that supposedly helped. The second fix was this major fix the devs kept talking about was coming. Announced as “big and difficult so it takes time.” etc. It ended up being generic low-poly models instead of being entirely invisible along with blob shadows to represent those that were completely invisible. And they’d pop in and out based on draw distance. It was ridiculous. They could have implemented the basic blob shadows months previous. My (large) guild simply ragequit.

        From your description it is still not fixed. It should not happen at all. WoW could send out the location data for 120 players with ease years earlier. Obviously these two games had different engine architecture. It was still unacceptable. Not when GW2 had these big battles and zone events as such a major feature of the game.

        1. Zagzag says:

          It works fine these days. There was a change to how nameplates work in PvP/WvW that presumably came after you’d already left, and that pretty much fixed any remaining problems but much too late for a lot of players.

          I don’t think it’s ever been sending location data that was the issue, but rather that most players’ hardware isn’t up to the task of rendering 120 player characters on screen at once, along with the required particle effects for all the skills they’re using. Early on they automatically rendered only a fixed number and the rest were invisible, now you have more control over how many should appear as high/low poly models and how many get nameplates only. The amount of bandwidth required to keep track of where those characters and standing and what skills they’re using shouldn’t be much of a problem in comparison.

          1. Steve C says:

            Sorry, no. There’s no way that is true. Rendering 120 things on screen at the same time is nothing. Particle effects are nothing. Again, WoW has been doing it for years.

            However player bandwidth is logarithmic or exponential problem as each player has to send all their relevant data, skill activation, etc to the server, the server has to crunch it and send it back out to the other 119 players + show the results to the player who did it. So 120 players interacting with each other is putting a load of 120^2. Which is like having 14400 individual players interacting with the server. Except probably doubled as there are both inputs and outputs. (Though I’m sure a programmer could come up with the math behind the actual problem.)

            “now you have more control over how many should appear as high/low poly models and how many get nameplates only. “

            That is not working fine. That is a kludge. A kludge that should have been immediately available. I don’t think you are getting me when I say THAT, that thing you just described there… THAT is not acceptable. When I said “not at all” I meant “not at all.” Not in the kind of game they made. Not after years of fixes. Not when other games (both newer and older) do not have to make that trade off. (Planetside2 comes to mind.)

            Like the point of Shamus’ blog post- There was something deep inside the GW2 engine that made it unsuitable for purpose. The wrong tool for the job.

  16. Abnaxis says:

    Instead, EA just told everyone to switch to Frostbite and left them to work out the details on their own

    From the Kotaku articles, my understanding is that EA has a Frostbite support center for their subsidiaries, which is supposed to help the with engine engineering issues. However, if you aren’t working for one of the tentpole properties any issues you have get moved into the back of the queue, behind FIFA and the linear shooters. So there is support for Frostbite on paper, but you have to play the corporate politics game to get it and you’re in a zero sum competition for it with the other subsidiaries

    When people talk about all the indecision and mismanagement from Bioware, I can’t help but wonder if indecision isn’t really the result of a corporate culture that yanks support away from you if you commit to a realistic design that doesn’t get support from the right higher up.

    1. Ninety-Three says:

      a corporate culture that yanks support away from you if you commit to a realistic design

      That might be a problem which exists at EA somewhere, but Schreier’s Bioware articles described the exact opposite of committing to a design.

      1. Nessus says:

        He’s saying that in a situation where committing to a “wrong” design will get you dropped down the support queue with potentially catastrophic consequences for your project, deliberate indecision can become a strategy. Stringing along the higher ups in charge of prioritizing support by vaguely implying you have a plan they’ll like rather than showing a plan they can make a definitive judgement on becomes a way of keeping your current level of support for as long as possible. This was one of the conditions implied to be at work by the Kotaku article about Anthem’s development.

        If true, this would make the execs who choose what project gets Frostbite support partially culpable, as they created a system where consequences are so severe and unpredictable that the mere threat of them is enough to tank productivity almost if not as badly as said consequences would themselves.

        Though if Bioware wasn’t in someway strong armed into using Frostbite, they could have avoided the problem entirely by using Unreal. As commenters above have noted, even if Unreal would have been just as much work as Frostbite, since Unreal has a massive support base and most modern devs are virtually guaranteed to have some degree of training and/or experience with it already.

  17. Raygereio says:

    Frostbite somehow being evil incarnate is often repeated, but honestly that screams of apologizing the developer’s screw ups. It’s honestly a bit baffling to me that so many people keep pointing the finger at Frostbite, when DA:Inquisition came out before and had none of the big problems ME:A had. And the idea that “Frostbite isn’t suited for RPGs because it’s a shooter engine” is especially silly when the Unreal Engine is part of the conversation.
    I mean, sure the use of Frostbite was undoubtedly a mandate from EA. But you work with the tools you have. And sure, having to use Frostbite did present certain issues and hurdles to overcome. But if Bioware had taken their Eclipse engine and build a new engine on top of that, they would have different issues and hurdles to overcome. That’s just part of game development.

    Basically: Frostbite wasn’t the reason ME:A sucked. Nor is the reason Anthem sucks. Bioware today has a massive problem at a project management level. Using different tools won’t fix that.

    Fun bit of trivia:
    Dragon Age: Origins was originally developed on a heavily modified Aurora engine (KotOR1, Jade Empire). The Eclipse engine meanwhile was a sepeate project for a new, open world engine. When DA:O was moved over to the Eclipse engine, they found that the whole open world thing didn’t work with DA:O’s concept. So it was all scrapped.
    I can’t help but wonder what sort of game a streaming, open-world Dragon Age Origins would have been. Or how that would have shaped the cRPG market if it was a success before Skyrim.

    1. Trevor says:

      I know this really isn’t this site’s genre, but Madden and FIFA adopted the new engine seamlessly. They also faced a change of genre from Frostbite’s cinematic shooter to sports game and while there are some similarities in terms of closed worlds and very defined playing areas there are also a good amount of differences.

      Not only were they able to transition from one engine to the next, they did so seamlessly. People (rightly) comment that Madden/FIFA Year X is identical to Madden/FIFA Year X-1, but the Frostbite Madden games play exactly the same as the Ignite Madden games. If the EA Sports people were able to figure out the transition, and again, do it so smoothly that you don’t notice it in gameplay, you only see improved graphics, I’m inclined to put more of the blame on the people in the BioWare studios.

      1. shoeboxjeddy says:

        VERY important note about FIFA/Madden. According to all the devs at Bioware interviewed for the stories about what happened to Andromeda and Anthem, the reason the sports games transitioned so gracefully to the new engine is that they are the big money makers for the company and got all the tech support and TLC on the engine they asked for. Bioware got to take a number and wait.

    2. Gurgl says:

      It doubtful it would have been very good, there is nothing in how Dragon Age Origins is structured that would noticeably improve with a streaming world.

      It’s a big deal in Skyrim because you are in first-person, so “if you can see it in the distance, you can go there” has a lot of punch especially as a solo hiker with a real-time action combat system. It just had to feel adventurous and exploration-centric like this, it’s the natural conclusion of everything they tried to do in the previous Elder Scrolls games.

      Meanwhile, Dragon Age Origins was about its pausable party-based combat system that encouraged seeing the battlefield from above, and quets / dialogues, none of which would be better with an open world. It too was the natural conclusion of their previous RPGs (barring multiplayer).

    3. guy says:

      I’ve seen an Inquisition postmortem that reported that they struggled to implement features, especially leveling, in Frostbite and it cut into their dev time and contributed to its initial technical issues. Of course, Andromeda came out after they did that, but for some reason they made a totally separate fork from the DA:I one and had to implement leveling from scratch.

  18. OldOak says:

    To come first with some biases I already have about the BioWare Frostbite games:
    – They look gorgeous (honestly, I like the art they could deliver); yes, open world wise they might improve on the LOD (quite rough sometimes)
    – I like Dragon Age: Inquisition’s character movement/control approach; took a while to get used with, but it becomes second nature (at least on PC). While I’m not too much in the lore of medieval fantasies, this is the one that really kicks
    – I’m not one of these haters of Mass Effect: Andromeda; I also can pinpoint the things that I clearly dislike, the obvious bugs left
    – Anthem (No. Not too much into MMO here, sorry); but still the art looks quite mature.

    I think the real thing with the switch to Frostbite (regardless on where the initiative – hah! – came from), was prioritization.
    In Kotaku’s article there is a small mention that with Anthem, the “big boss” Frostbite went twice in Edmonton for demos (and my bet is not only for demos). So in Frostbite HQ there was some knowledge that something is not right with the usage of the engine with BioWare games.
    But I’d guess the main client for customizing Forstbite were the EA sports franchises, everything else, lower class citizens. I’d expect (like in any software environment) there were requests for enhancement placed by the brood of developers that were always pushed to the end of the queue.
    From what I could gather (and see — hell, I installed Battlefield IV on my computer just to get a glimpse of Frostbite in its native gameplay to compare it with Dragon Age: Inquisition), the real main complain from BioWare team was with tooling. Without too many details provided, and trying to read between the lines, it’s mainly that there are not very easy “in-engine” means to do testing, but you’ll need to “build”. I guess this is because (as repeatedly stated) they might have some “in-engine” tooling that only applies to the first person shooters. Anything else seems to be (maybe local/in-house at the other studios) “add-ons” that require the whole build process to test.
    This would definitely make sense. I’d say that due to the lack of “RPG frameworks” in Frostbite, and the need for the development team to code them, this is was the main drawback BioWare had. What the FIFA/NBA & co. got handed over nicely wrapped up and doing Next/Next/[…]/Finish to install, they had to do on their own. And of course, the lack of appropriate documentation (“why would you need any when you have the source code available” <- when you hear this with a large project, you usually know it means trouble).
    This really means that while difficult to properly allocate the blame, in an ideal business environment BioWare should've had to have the means to properly get the support they needed for their engine changes early in the development cycle.

  19. Agammamon says:

    The way to properly handle something like this would be to create a specialized engine team that was just focused on building, extending, documenting, and supporting Frostbite for EA’s subsidiaries.

    If EA had been smart they could have spin-off the Frostbite development team into their own studio and start marketing it *to other developers* ala Unreal, Unity, or Gamebryo. You get a team of dedicated people developing the thing, all the in-house tweaks the EA development studios figure out can be put back into the core product as new features, and you can help offset the costs by licensing it to third party developers. Which, if nothing else, takes some money *away* from Epic (if you want to get nasty and are just trying to hurt).

  20. Sartharina says:

    You’re off-base in two places. It’s already been covered how they use Eclipse, not Unreal… which I think was ultimately shuttered due to the proprietary nature of the engine. Essentially, Bioware COULDN’T use Unreal, given the chunk of profit Epic wants to take from major games developed on it. So, they had to use an in-house engine: Eclipse, or Frostbite.

    The bigger issue, though, is how you sell Frostbite short. Frostbite isn’t merely for “Linear shooters”. It’s the engine used for the Battlefield games, which are massive multiplayer games with HUGE battlefield maps capable of seamlessly transitioning between sluggish foot and high-speed vehicle operation (Which, as far as the engine is concern, is largely just camera transitions). But it does both open world and multiplayer very well.

    1. Agammamon says:

      Unreal Engine 4 is now free to download, and all future updates will also be free. Developers of commercially released games or applications will pay Epic a 5 percent royalty on gross revenue above $3,000 per product, per quarter.

      https://www.polygon.com/2015/3/2/8134425/unreal-engine-4-free-epic-games

      I’m not seeing the huge chunk of the profit that Epic is taking.

      1. Nessus says:

        Moreover, even if it was a huge chunk, sharing the profits from a successful game is better than keeping all the profits from a failed game.

        Honestly, if the Kotaku article was accurate, there were several points in Anthem’s development where it seems like a cost analysis should have mandated a switch to a better supported engine, even if it meant restarting almost from scratch and taking on a licensing fee. If you’re gonna spend the next few years working on the project anyway, better to spend those actually getting shit done rather than faffing about because management needs to keep politiking in order to maintain a space in the corporate support queue.

        If you believe in your company enough to unironically handwave serious concerns with “Bioware Magic”, that should mean you believe enough to be comfortable with the financial risk of licensing an engine.

      2. Richard says:

        That “5% revenue” is what killed Unreal for us.

        If you’re selling actual hardware that’s running the software, then 5% of gross revenue can easily be half your profit margin.

        So in some markets it is a huge chunk. Not for games though.

        1. Agammamon says:

          So, you were never going to release your game through Steam either? With their 30% default bite of the gross? Or are you saying that that 30% was acceptable but the 5% wasn’t?

          What are you using in place of Unreal then? An in-house engine? Will the development costs of that be less than 5% of your gross revenue?

          Or did you find licensing with another third party developer under more favorable terms? And if you did, why couldn’t Bioware?

          1. Richard says:

            Other engines charge per developer seat per year of development, or a fixed charge per year.
            We chose one of them.

            It is interesting how you didn’t read any of my comment at all.
            Steam only sell their own hardware, plus the HTC Vive. I think it is very safe to say that Valve don’t get a 30% cut of the Vive.

            If you pay $100,000 for a piece of industrial machinery (let’s say it’s a car), then the dealer who made the sale probably takes 20%-25% of that, with the manufacturer taking the rest.
            (Note that the list price is likely to be $120,000)

            So the manufacturer gets $75,000 of gross revenue. 5% of that would be $3750, per item sold.
            Manufacturing margins tend to be around 10%, so that single licence just ate half your profit.

            Valve charges something like $20,000 to $40,000 for Source depending on negotiations, Unity is $1500pa/developer.
            So the Source engine is cheaper at 11 idividual sales, assuming the top end (as it won’t be on Steam).

  21. Solism says:

    Hi Shamus,

    Concerning footnote 1, here’s the source.

    And a quote for the relevant bit:

    Three people who worked on the game point to one critical moment early in pre-production, when BioWare decided to switch animation programs from 3D Studio Max to Maya, as a move that cost technical animators a great deal of work. Although it was seen as a necessary overhaul by BioWare’s leadership—Autodesk, the company behind both 3DS Max and Maya, was recommending that game developers shift gears—the animators were upset that it happened during pre-production rather than before they’d started working at all. “All that technology was invalid, simply because we’d used a red pen instead of a blue pen,” said one developer, lamenting the months of progress they lost. (Later, some of Andromeda’s animators wound up using 3DS Max for a large chunk of their work anyway.)

    “That was hell for all the animation team,” said one lead developer. “That, mixed with the facial animation tech that came late, you had all the conditions to just have everything go really badly from an animation point of view.”

    So as Mass Effect: Andromeda entered production, animation became a big red flag. “You have to lock all your tech in pre-production to move into production,” said a developer. “One that was not locked was facial animation. It was flagged in 2014 as one of the major risks of the project.”

  22. Smejki says:

    I don’t know but all the “zomfg frostbite is to be blamed” smells more and more of bullshit to me the longer they try to play this card. Now, I know taking a half-baked engine and expand its feature is a huge undertaking but I also know that EA supposedly has the best engineers and a lot of money and therefore time. And I also know we bought CryEngine – a thing with semi-open world capabilities with zero to limited support for melee combat, dialog systems, inventory systems, massive-scale entity persistence, quadrupeds, large navmesh, clothing systems, AI behavioral trees, quest systems, and more than 10 characters in the scene. And we, with our extremely limited programming staff and resources, turned it into something that is almost 100% where we wanted it to be and made a fucking high-end open-world RPG with it – effectively in 4 years. The only things we didn’t have to improve too much was the lighting engine, the physics engine, animations systems, and vegetation handling. Even the timeOfDay simulation was a bit lackluster as it only rotated the Sun in a simple circle which meant 12 hours daytime and 12 hours of night time which, for a game set in Central European summer, is unacceptable. We even developed our own scripting language. And don’t get me started about the tools we had to develop.

    I could understand if their first Frostbite game sucked. But I cannot accept a third failure.
    All I can see is some permanent urge to reinvent the wheel every time, or long periods of indecision leading to half-baked tacked-on last-minute solutions that have to be scrapped next time, zero planning for the needs of future projects, bonkers tech-versioning system etc., ie. incompetent management all over again. Or they truly have very incompetent programmers which I refuse to believe.

  23. Majromax says:

    I can’t prove that Frostbite was mandated by EA,

    You don’t even need a management diktat to use Frostbite. All you need are a few nudges and economics.

    All EA management needed to say was ‘Frostbite is free, including engine support, and we won’t give you additional funding for a third-party engine license.’

    The fallout is encapsulated in this quote from the Schrier Anthem article:

    “Frostbite is like an in-house engine with all the problems that entails—it’s poorly documented, hacked together, and so on—with all the problems of an externally sourced engine,” said one former BioWare employee.

    EA may have treated Frostbite like a truly “free” engine. Replicating all of the classic problems of bad customer service, support was a cost to be minimized. For an externally-licensed engine, support contracts are a revenue source; for a truly in-house engine programmers know who to yell at over the lunch hour.

    The cobbler’s children have no shoes, only it’s a software conglomerate with a game engine.

    1. Nessus says:

      A big plus with a publicly available engine with a wide userbase like Unreal is that even if you can’t get official support on the horn for whatever reason, a quick google search will still bring up gaggles of forum posts by other Unreal users trying to do the same or similar things that can at least get you a good head start, if not outright solve your issue.

      The support nightmare described in the article flat out wouldn’t have happened with Unreal. I can’t help but wonder how the cost of wasting paid man-hours forcing a whole dev team to faff and stall for literal years actually compares to the licensing cost for Unreal. What percentage of the game’s budget was just poured down the toilet in the effort to avoid licensing?

      1. Majromax says:

        a quick google search will still bring up gaggles of forum posts by other Unreal users trying to do the same or similar things that can at least get you a good head start, if not outright solve your issue.

        That’s true. I wonder what the dollar value of foregone support-via-Stack-Overflow would amount to.

        I can see why a beancounter would think the decision makes sense, but overall it just tells me that EA is too big: its studios aren’t acting as coherent parts of a bigger whole, they’re acting as independent operators with an agreement to not compete with each other in the market.

        I can’t help but wonder how the cost of wasting paid man-hours forcing a whole dev team to faff and stall for literal years actually compares to the licensing cost for Unreal.

        To be fair, that also comes down to seemingly bad project management on Bioware’s part, without any assistance from EA. No matter the exact bad hand dealt by EA executives, Bioware did little to mitigate the damage with either Andromeda or Anthem.

  24. Alan says:

    I knew a guy who worked at EA back in the late 90s who had a similar story. It’s been almost 20 years, so some details may be wrong, but as best I remember it:

    Some shooter, probably a Call of Duty game, was getting a new release. The old engine was deemed no longer good enough. So the team was told to move to the most realistic engine EA had, one whose realism earned solid reviews. They were told to implement an FPS using the Madden engine.

    I’m a bit boggled someone thought this was a good idea. The problem spaces are SO different.

    Apparently in practice this meant writing a new engine on top of the gutted shell of Madden.

  25. I think it’s a shame they ditched the Eclipse engine (awesome for RPGs), they could have refactored that and kept dev tools similar enough, and more importantly more mod friendly as well.

    Dragon Age 2 was the last game to use Bioware’s “old” Eclipse engine (that has a legacy back to KoTOR and Never Winter).

    It’s funny that people shout at Bethesda to scrap their engine, yet seems to cry that Bioware should have stayed with theirs. (I guess some has a opposite view to this too though).

    Myself I don’t have an issue with Eclipse nor Creation Engine, as a programmer I know you can refactor everything, and if you make the engine modular enough you can replace the engine in chunks over the years, allowing a slow migration.

    I’m certain that open world and asset streaming could have been added to the Eclipse engine, and I doubt the cost would have been higher than going all in on Frostbite.

    Here’s the odd thing though, a updated Eclipse engine would probably have worked better with Andromeda (each planet would be a area/map, actually you only moved around a small part of the planet) than Frostbite.
    One thing Frostbite is better at than Eclipse engine is the rendering obviously.

    Maybe a future Mass Effect can be better by adding back more or the RPGness to it rather than FPS stuff, I’m not saying pausable KoTOR like combat, but something in-between (and maybe without a voiced player character?).

    Another issue with EA using Frostbite as “the engine” is that only DICE are developing it, it would be better if they had a internal git that all studios could contribute to.
    Also, IMO they should have had three engines, a FPS engine for high framerate twitch gaming, a RPG engine (Dragon Age etc), and a mmo engine (for open world/multiplayer/mmo games).
    And all three with streaming asset abilities obviously.

  26. Brian N. says:

    The idea that EA has a hands-off approach and has for years is belied by the documented cases of them meddling projects, franchises, and studios to death (Visceral counts for all three) and the consistency of the narrative is what worries me. It reeks of, “you’ll fall, and you’ll do it exactly how we tell you.”

Thanks for joining the discussion. Be nice, don't post angry, and enjoy yourself. This is supposed to be fun. Your email address will not be published. Required fields are marked*

You can enclose spoilers in <strike> tags like so:
<strike>Darth Vader is Luke's father!</strike>

You can make things italics like this:
Can you imagine having Darth Vader as your <i>father</i>?

You can make things bold like this:
I'm <b>very</b> glad Darth Vader isn't my father.

You can make links like this:
I'm reading about <a href="http://en.wikipedia.org/wiki/Darth_Vader">Darth Vader</a> on Wikipedia!

You can quote someone like this:
Darth Vader said <blockquote>Luke, I am your father.</blockquote>

Leave a Reply

Your email address will not be published. Required fields are marked *