I just remembered Epic made a new physics engine for UE4 for destructability
-
Ever wanted an RSS feed of all your favorite gaming news sites? Go check out our new Gaming Headlines feed! Read more about it here.
-
We have made minor adjustments to how the search bar works on ResetEra. You can read about the changes here.
[Gamesindustry.biz] Former Bioware GM opens up about difficulties of Frostbite engine
- Thread starter dex3108
- Start date
I mean I get this. But like...DA:I was on Frostbite and it was fine, and the games only got worse from there.
So even if it's hard to deal with, the fact that two games later are still a programming and scripting nightmare doesn't make me feel sympathy, on top of the previous directors switching because they knew they'd have to eventually.
I get that it's not "made for" RPGs, but every game since DA:I has removed and deleted features and content, and I just can't feel sorry for a team getting worse in an engine they should be getting better with after 5+ years.
I feel sorry for the devs for having a difficult engine to work with, but this sounds like either Frostbite is terribly managed, Bioware is terribly managed, the resources all left the company, or a mixture of it all. And even that doesn't fix the problems with things like Anthem.
So even if it's hard to deal with, the fact that two games later are still a programming and scripting nightmare doesn't make me feel sympathy, on top of the previous directors switching because they knew they'd have to eventually.
I get that it's not "made for" RPGs, but every game since DA:I has removed and deleted features and content, and I just can't feel sorry for a team getting worse in an engine they should be getting better with after 5+ years.
I feel sorry for the devs for having a difficult engine to work with, but this sounds like either Frostbite is terribly managed, Bioware is terribly managed, the resources all left the company, or a mixture of it all. And even that doesn't fix the problems with things like Anthem.
But why tinker with a Toyota when you got a Ferrari available?
At least it's functional.
Bethesda keeps trying to supercharge their Subaru when it will always be a Subaru no matter the mods.
At this point, I would guess that it would not make sense to move away from Frostbite. All the issues and development pains have surely lead to a lot of new tools, upgrades, knowledge, and experience that should pay off with each new game. I would be surprised if the new Mass Effect and Dragon Age games suffer from the same engine struggles that were cited for ME:A and DAI.
They are not stupid. The idea of one engine that get improved by multiple studio with shared knowledge among all your dev is a very appealing concept even without thinking about money. It's every programmer wet dreams. Even ubisoft was thinking of doing the same at the time instead of having multiple engine that do very similar things. Turn's out having one super engine that does everything just isn't very practical after all, at least not the way they have done it
DA:I and Andromeda weren't made by the same team, so they are not getting worst at it, your comparing a team of veteran vs a studio that had never shipped a single game... And pipeline content definitely matter even if the game run and looks good. Iteration is a core concept of video game dev and affect the quality of every single aspect of a game
That begin said Bioware poor management probably didn't help either, if you have incredibly short deadline because of poor decision, having a cumbersome engine is just going to make things even worse.
Last edited:
having one engine that can do everything only seems to work when it can't do everything at the best possible performance. hence why Unity and UE can made damn near every genre, but doesn't come close to bespoke engines out of the box
DICE are a talented team that built Frostbite specifically for their needs and know how to use it. Them moving to UE4 would take away the advantages that they have in a custom internal engine designed specifically for their games with high performance. DICE game devs can communicate quickly and easily with the engine devs, so that actually makes their development more quick and efficient compared to having to wait for Epic support.
Also DICE games don't really lack in content, especially how fast they have to churn out new games for EA almost every year.
EA should have been more like Ubisoft and Sony, each dev team either choosing third-party engine or having their own engine for their own needs.
Nope, EA invested too much on Frostbite at this point, I remember the Need for Speed games being hell to put together till Ghost got a handle on it.
However, some games still use fucking Ignite.
Frostbite needs to die a fire. It's so obvious Bioware and ever other team has buckled under its inefficient weight. Speaking the truth about Frostbite (or any other subject for that matter) should not be met with resistance. I can't wait till they ditch the anathema to their development. I want to love an EA game again; look Jedi Fallen Order doesn't use Frostbite (though it needs its own optimizations and was rushed obviously but not at the fault of UE4).
I still think the engine is a scapegoat for more systemic issues that they were always able to out-develop/out-crunch until they had to learn a new engine and make it do things that weren't first person shooters. From what I've read, they are essentially developing the entire game within a year of release because core gameplay elements, narrative, and everything else are in a flux until, oh shit, we have to get this thing out in a year. Now your buffer to fiddle around, learn, and get support for the engine is gone and everything hits the fan. Learning new tech is just part of being a developer but you need time to do it. Time you don't have if you spend 6 years deciding if you should add flying or not.
Any interview with any BioWare employee usually goes:
These last two years of Madden have been the worst in franchise history.
I know he/she is just trying to be respectful to the Dice guys who built the engine, but it just sounds like the tool-chain wasn't up to the job.
The point of having a comprehensive middle-ware engine is that the tools available make artists, animators, programmers and game/level designer's lives easier.
You ca have all the most bleeding edge rendering tech in the world included, but if the folks working with the tools can't create or move assets through the dev pipeline and iterate efficiently enough, then the whole thing is just useless.
I thought EA had a central technology group that maintains Frostbite and supports the other EA devs using the engine? Perhaps EA needs to commission them to spend more time developing the engine toolchain before forcing the tech onto all their in-house devs (just to save a few bucks on Unreal royalties).
What's more valuable to you, EA? The future of your previously multi-million-selling IP like Mass Effect, or saving the cost of paying Epic to use UE4?
There's probably some truth there, but it doesn't explain why other EA franchise seem to suffer from problem they did not have before. Even EA biggest moneymaker, that can get as much support as they want like Fifa and Madden seems to be affected (I don't play either of those so I have no personal experience there)
The idea of a single multipurpose engine goes way way beyond just saving on royalties (as I said earlier in the thread)
They made every studio use the engine to improve it. Every studio was supposed to contribute to it. As per a Frostbit programmer they purposefully prioritize power over simplicity, which is fine when you are the one that as to deal with said engine afterward and not so much for other studios/dev.
Last edited:
I wish they weren't so general about it...
"Lets you do nifty things!"
like what exactly?
and was it worth the upkeep?
Every decision they make is about the here and now.
"Lets you do nifty things!"
like what exactly?
and was it worth the upkeep?
EA is always in it for the quick buck...
Every decision they make is about the here and now.
They had 8 years using that engine to build the tools they needed to make their games and yet somehow ran in to the same problems each and every game. It's a monumental management failure and it sounds like he was one of those responsible.
Ha, no, that quote wasn't from him, but Flynn was very open about Frostbite's difficulties in my book.
Pathetic post, it doesnt compare in any way
BF 1 - V absolutely are destroying anything Capcom ever did. V even more so, please call us when RE engine will support ray tracing at this scope
Says it doesn't compare in any way.
Still go and compares then.
Because its possible if they use another engine it might actually be cheaper and faster to develop a game. I mean it depends on how difficult it is to retrain and lots of factors. You have to pay people to develop that engine, keep it up to date and also support your other developers using that engine, and if it takes the developers a long time to develop in it, and they are always having problems, then it might be costing you more to use your bespoke engine than it would to use another engine like Unity or UE4.
It could. But you'd need a lot of expertise with that engine too.
I can think of only one game developer outside of EpIc that could do it and that would be The Coalition.
Honestly they should stick with Frostbite. But they should expand the toolset to effectively be an "internal" UE or Unity. At the cost of performance. That would make teams much more efficient.
Its either that, or build large teams in each studio dedicated to frostbite.
Isn't it mostly the toolset and the engine's original lack of modularity/rigidness that's the issue? It was meant for FPS games, and not designed for generalized use outside of that. I recall reading in one of these postmortems that EA set up a DICE team responsible for engine support within the company, but there's only so much you can do without creating a brand new engine outright to fix some of these issues and they're constantly busy so getting access to them isn't a given
Good luck, another game where they will have to spend 3 years fucking with the engine to actually make the game in 8 months.
BioWare needs to just use UE4
I have heard, like 5th hand, that the Coalition rewrote large swaths of UE4 to get it to do what it does in Gears 5. It was no small feat.
Not that BioWare didn't have to write a bunch of core engine stuff for DAI (and Anthem), but UE4 isn't just a magic bullet.
The Coalition worked very closely with Epic. much like Lionhead, their changes probably got rolled into the public build
The Coalition also has the benefit of being in XGS along with other studios using UE4 like Rare so im sure they share tech and technique with eachother on top of with Epic. A lot of those changes though also probably saw them get folded into UE4 proper
the majority of Frostbite wizards are working on (or worked on) Battlefield. That doesn't make it an intrinsically bad engine (or set of tools).
A hammer might be shit at painting walls, but its really good for nailing nails. And its easy to see how it could be hard Especially when you've never nailed a nail or used a hammer to paint a wall. And all the hammer-nail specialists are working on another project across town. So you're stuck learning how to use a hammer to paint walls when corporate should have just given you a fucking paint brush
A hammer might be shit at painting walls, but its really good for nailing nails. And its easy to see how it could be hard Especially when you've never nailed a nail or used a hammer to paint a wall. And all the hammer-nail specialists are working on another project across town. So you're stuck learning how to use a hammer to paint walls when corporate should have just given you a fucking paint brush
I think Bioware said that Frostbite works best for shooters and racing games (since vehicles are important for Battlefield), and Need for Speed hasn't really had any engine trouble
for shooters for sure but honestly even some of the vehicles in Battlefield are a little wonky. The jeeps/bikes have never felt that great imo (tanks and planes tho are always fun).
Cant speak too much for the newer NFS as I havent played em too much but I will say from what I have played games like Forza Horizon are leagues ahead.
Did changes/improvements get rolled back into the engine? Almost undoubtedly, however, usually when you're seeing the best of something, it's due to specialization. ie- the improvements made to the engine to make Gears 5 perform like it does are things that are mostly just useful to make Gears 5 (or something exceptionally similar) perform like it does, rather than generalist improvements that would apply to many other games using the engine.
Basically, what I'm saying is: an off the shelf engine can make developing waaaaaaayyyy easier than going without one, but usually when you see best-in-class anything, it's specialized.
well of course when you see games like Gears 5 or Sea of Thieves they have 100% been messed around with to fit the the studios need. That being said I think the value of having something more "standard" or easy to use at its base can help push a game and team forward much more easily (and they will of course make changes to fit them too)
Frostbite can make some killer shooters but its clearly not easily stretched and suited for shooters, racers, sports, action/adventure, VR etc etc and needs I feel like sometimes to much work to try and force it into something kinda ok where you can get kinda ok from something like UE4 without the pain in the ass.
haven't all of the frostbite battlefield games had various bugs and issues that never really got fixed? maybe frostbite is just shit, even for DICE
Correction on my part: He made his comments before the Anthem stories got out. Wow.
From my understanding the issue with Unity has always been optimization. Idk how much that has improved over the last few years though.
The problem with engine programming is something that might sound simple in theory isn't always easy to implement in practice, due to how the engine pipeline prioritises tasks and the tools programmers have available to make changes. Every game requires amendments in every engine. It's the extensiveness of these amendments that vary, and the difficulty implementing them. If I recall correctly, Schreier's piece on Andromeda highlighted how BioWare had issues with Inquisition and Andromeda simply due to Frostbite's difficulty handling NPCs that can be triggered into conversational status and whatnot. The backend of the engine was just very, very difficult to deal with in building and implementing these features, whereas adding something like Andromeda's vehicle was much easier as large chunks of code and optimisation already existed within Frostbite.
In programming, sometimes adding what might seem like something absolutely basic doesn't always work out smoothly. Populating an area with blank slate NPCs that walk around a bit on schedules, and can be triggered into conversational status, might seem simple on paper but the actual execution could end up a bit of a nightmare. Simply due to how engines handle code pipelines. Maybe you implement this feature and the performance absolutely tanks. Maybe, for some reason, NPC pathfinding keeps bugging out. Maybe it keeps loading the wrong conversational trees for each NPC, or the engine crashes when it tries to recall the depth of which a conversation has already been explored. It's hard to know the specifics, but these can be problems.
On top of this, the tools and documentation available for an engine can make or break usability with programmers. If I recall correctly, this is exactly what stifled Crytek trying to push one of the iterations of CryEngine into the market, where compared to Unreal Engine 3 the document and support wasn't anywhere near as user friendly or robust. Meanwhile Epic made a name for themselves by building middleware, namely Unreal Engine 3, that specifically targeted and industry that was rapidly being faced with enormous expectations from the rendering pipeline. Is UE3 perfect? No, but it was swept up and adopted by so many developers because, as far as I'm aware, it had extensive, reliable support and documentation and (at least from my own experience) a constantly evolving and approachable user interface that made programming life easier.
What Flynn is highlighting here can be seen in the original trilogy. It's pretty evident as time progressed BioWare became more efficient and comfortable with Unreal Engine 3, hence sticking with it across three games. ME1 using an early build of the engine and reaching beyond its capabilities were surely a sticky learning curve, but ME2 and ME3 were right on the mark. By the end it really seemed like BioWare were able to do more or less exactly what they wanted with the engine, visually and mechanically, within the confines of their development scope.
Frostbite is the complete opposite. And I mean, obviously different strokes for different folks; maybe some people love Frostbite and feel BioWare's frustration with the engine is more demonstrative of their own managerial and technical incompetence than engine faults. But Flynn's frustrations with the engine aren't the first we've heard of it, and I'm gonna be honest, having my ear to the ol' journo grapevine, programmer frustration with Frostbite seems to be a poorly kept secret.
And I'm gonna be honest; I kinda feel even DICE is floundering with their own engine. Battlefield V launched a hot buggy mess and stayed that way for months and months, the UIs were (and still are to an extent) laggy and littered with streaming issues, vehicles feel weirdly floaty with poor physics, every patch seems to fix one thing and break another, and there's this lingering feeling that while a rushed production is the most likely culprit that Frostbite isn't helping matters.
So yeah. Frostbite Engine; insanely gorgeous and powerful and capable of utterly incredibly looking stuff, but maybe not a great engine to work with, ranging up to a nightmare if you're BioWare.
In programming, sometimes adding what might seem like something absolutely basic doesn't always work out smoothly. Populating an area with blank slate NPCs that walk around a bit on schedules, and can be triggered into conversational status, might seem simple on paper but the actual execution could end up a bit of a nightmare. Simply due to how engines handle code pipelines. Maybe you implement this feature and the performance absolutely tanks. Maybe, for some reason, NPC pathfinding keeps bugging out. Maybe it keeps loading the wrong conversational trees for each NPC, or the engine crashes when it tries to recall the depth of which a conversation has already been explored. It's hard to know the specifics, but these can be problems.
On top of this, the tools and documentation available for an engine can make or break usability with programmers. If I recall correctly, this is exactly what stifled Crytek trying to push one of the iterations of CryEngine into the market, where compared to Unreal Engine 3 the document and support wasn't anywhere near as user friendly or robust. Meanwhile Epic made a name for themselves by building middleware, namely Unreal Engine 3, that specifically targeted and industry that was rapidly being faced with enormous expectations from the rendering pipeline. Is UE3 perfect? No, but it was swept up and adopted by so many developers because, as far as I'm aware, it had extensive, reliable support and documentation and (at least from my own experience) a constantly evolving and approachable user interface that made programming life easier.
What Flynn is highlighting here can be seen in the original trilogy. It's pretty evident as time progressed BioWare became more efficient and comfortable with Unreal Engine 3, hence sticking with it across three games. ME1 using an early build of the engine and reaching beyond its capabilities were surely a sticky learning curve, but ME2 and ME3 were right on the mark. By the end it really seemed like BioWare were able to do more or less exactly what they wanted with the engine, visually and mechanically, within the confines of their development scope.
Frostbite is the complete opposite. And I mean, obviously different strokes for different folks; maybe some people love Frostbite and feel BioWare's frustration with the engine is more demonstrative of their own managerial and technical incompetence than engine faults. But Flynn's frustrations with the engine aren't the first we've heard of it, and I'm gonna be honest, having my ear to the ol' journo grapevine, programmer frustration with Frostbite seems to be a poorly kept secret.
And I'm gonna be honest; I kinda feel even DICE is floundering with their own engine. Battlefield V launched a hot buggy mess and stayed that way for months and months, the UIs were (and still are to an extent) laggy and littered with streaming issues, vehicles feel weirdly floaty with poor physics, every patch seems to fix one thing and break another, and there's this lingering feeling that while a rushed production is the most likely culprit that Frostbite isn't helping matters.
So yeah. Frostbite Engine; insanely gorgeous and powerful and capable of utterly incredibly looking stuff, but maybe not a great engine to work with, ranging up to a nightmare if you're BioWare.
yea, that's what I touched on earlier in the thread. it's been the same with all third party engines. especially with Unity prior to 5. hell, it's the same with internal engines. who knows how Nintendo's stack evolved since EPD makes all sorts of games
I'm wondering why Cry Engine is that much overlooked, Frostbite aside. Like, one of the early iterations literally handled a mmo, it should be able to do single player rpgs while delivering high quality everything
Frostbite is a great Engine for Dice and that is not surprising since they created it for themselves. The fault was EA then deciding what DICE made was good enough for everyone else and mandated its use. Frankly if I was Bioware I would be protesting until I got to use UE4 or another engine. Even 5-6 years ago there was people within EA speaking out about how bad the frostbite situation was in the fact EA understaffed the development team and the support staff for developers to reach out too
I've hardly heard of any really good dev experiences with CryEngine, aside from Crytek themselves (IE most studios I know that used to ship a game with it, would quickly change to something else for their next project). Not having any direct experience I only tidbits of experiences, but it certainly doesn't have the same resources behind it as other middleware these days and when visual quality is the number one marketing bullet point, it's easy to guess what pitfalls that may mean in terms of actual workflow.
Maybe has more to do with us making 5 AAA games in 4 years. :p