The pros and cons of UT developing engine software without in-house games

Unity no longer makes games. Their forays into game development, like the UT-associated Global Conflicts: Palestine and the Over the Edge Entertainment's GooBall, were spearheaded by Nicholas Francis, who has since left UT to pursue game dev independently. In a recent newsletter, UT touted the fact that they didn't make games as an advantage—because they could instead focus on making the best tools for developers. I'm curious what you guys think about whether this is indeed an advantage.

On the surface, it feels like a contradiction, but perhaps skepticism is just the inevitable stance of a game developer on this topic. The "eat your own dogfood" argument goes that if Unity were involved in game development parallel to engine development, the pragmatic needs of their game would:

1) Inform the direction of development on the game engine.

2) Provide increased pressure to fix low-priority but abrasive bugs, especially those of long standing.

3) Incentivize more rapid development in general.

4) Engender trust among customers who feel their situation is familiar to the corporation's.

5) Put a face on the engine's capabilities, ameliorating the age-old question, "which good games are actually made with Unity?"

Unity's disinclination towards in-house game development stands in stark contrast to its competitors. Epic and its subsidiaries have developed a number of famous properties, and their expertise in content production stands out when comparing the two company's tech demos and community initiatives (to wit, Epic's choice to involve the community in Unreal Tournament 4 post production).

However, UT is a savvy company. Their movement into the free development space showed both bravery and acumen, and their roster unquestionably boasts many talented decision makers alongside designers and developers. Unity began as a Mac-only product clearly inspired by Apple's ideal: render the complex into the simple and the essential. It's in the name. Likewise, UT's business practices are reminiscent of Apple's: often secretive; at their best—innovative. The object / component model is unquestionably weird. But there's also no question that it makes prototyping a cinch... even a joy.

So giving the benefit of the doubt to UT's choices, why might eschewing in-house game development be the right choice?

1) Close links to a specific game can create an engine best suited to specific gameplay formats and technologies. UE4 resists this accusation with counterexamples, but from what I understand the criticism still holds.

2) As we know, game development is very risky, AAA development especially so. Perhaps UT simply don't want to have to take on the risk of hundreds of artists, audio crew, designers, producers, etc. etc. for the esoteric benefits listed above? Just look at how Crytek's fortunes have risen and fallen again on the back of their game development efforts.

3) If UT keeps very close ties with specific developers, it might render in-house feedback on the engine moot. They would get the same feedback without the risk.

4) UT has stated that as a company they are simply more interested in making tools than in games. On the one hand, this sits uneasy because as one among a community of game developers the natural reaction is, "what's wrong with you guys?" On the other hand, could it just be as simple as that? A personal preference guiding a large, increasingly impersonal company?

So what do you think? Does UT's reticence to develop games with their own game development tools suggest a lack of confidence that they can "put their money where their mouth is"? Or does it show a conservative wisdom and the vision it takes to walk one's own path and not another's?

 

When I look at things like the UI/UX of Mechanic, all I can think is "damn, I wish these tools were made by someone who'd actually tried to make a game with them".

 

I think we discussed this a lot back when Epic got involved in the fray.

1) You can tell where Epics strengths are, no limitations of types of games but certainly it caters to the PC / MAC / Console market much better than it does mobile, it seems to me they were looking to leverage the latest Tegra and Metal tech and anything before that is a bit "whatever".

2) You don't really need that many people for AAA tech demos, I've seen really small teams and one man bands do things on par with the epic engine tech demos. It's because the engine developers make it easy to do so, that comes from years of experience making them types of games and knowing how to get around limitations effectively.

3) From what I've heard Unity has plenty of AAA engineers in there grasp, so they have the knowledge pool collectively. Nothing stopping them.

4) Last time I checked Unity was the only engine that supported nearly every platform available, I suppose you have to pick your battles in areas and of course there will be some trade off for doing so. I suppose it comes down to how many people care about said platform, I know of a few who develop for Vita and WII but I'd guess they are a minority..

What do I think?

If Unity could have a crack at a worst case scenario, which would be something like Witcher 3 off the top of my head and see how far they could get in Unity that would be very interesting. The amount of upgrades that would come off the back of an experience like that would be crazy. They should do it just as a learning experience..

 

I wish they had an internal game development team. I agree if they actively used their tools they would have clear, firsthand knowledge of how to improve their products. They could also just make some larger scale tutorial projects. Like with the 2D for example instead of the very simple 2D game do a real 2D scrolling platformer using tiles for background scenery (and even foreground for some parallax scrolling). Have enemies that are more involved. Only by moving away from overly simplistic demo projects can they start to run into the issues experienced by their users.

 

I think that Unity would benefit from having in-house game team working on actual game project. It's so obvious when using Unity for more complex stuff than simple "engine developer test case for specific engine subsystem". Good example was Mecanim when it was first released, yeah it was great to drag in some animations and to see wow, it's blending based on some input variable, but when you try to use it for more complex game situations, it was lacking a lot.

 

There comes a point in most software companies when you need to have a UX team that interfaces with key clients to find out what is needed. In house dog food eating only works for so long. Similar to making a game, it is best if you don't test your own work.

 

If I had to guess, I would bet the asset store would benefit more than the engine. I could see there being some little improvements here and there, but the major products would be tools that just end up on the asset store because they aren't ideal, but work fine for it's specific usage.

 

This topic comes up quite often and done so with the assumption that we have no internal teams working on making things with Unity. We in fact have 3 teams that create content for internal and external use with Unity, all with different goals. We also have a close working relationships with our premium and enterprise support customers, we get a lot of feedback from them, it also stemmed our sustained engineering stuff. Additionally there's the closed Alpha and Beta groups and the community site of course. We're a bit spoiled when it comes to feedback from real projects in development

That's what we've got currently, I'm not saying we would never have our own games studio, you never know, things may change, but right now we're concentrating a lot of our resources on shipping 4.6 and 5.0 and getting those as badass as possible. We can't deny the pro's but then there are also the con's, just because we've achieved in creating a successful and popular game engine doesn't mean we have the perfect formula to create an amazing and successful game.

 

Those teams sound interesting, @Aurore . By Unity-produced content, are we talking assets, tutorials, and demos? I was going to contrast these with full games, but for the purpose of improving the engine maybe the distinction is meaningless. For the purpose of prestige, it probably isn't!

Since you've distinguished between "creating a successful and popular game engine" and "creat[ing] an amazing and successful game," do you feel that these are not necessarily complementary goals?

 

All of the above, you're probably familiar with the content team's work in Learn, they create games and Assets for the purpose of teaching Unity http://unity3d.com/learn/resources/downloads

Our team in Stockholm push Unity's capabilities with their demos e.g. Doll demo, Viking Village http://blogs.unity3d.com/2014/09/18/global-illumination-in-unity-5/ Butterfly Effect http://unity3d.com/pages/butterfly

Then our team in Seattle make internal and Biz Dev stuff, soz no links.

I do think it would be cool if we made commercial games but I don't think it's necessary, we have a great relationship with this community and the teams above, this help us discover solve pain areas. All of you and our content teams work on a huge variety of games and projects, I think this helps keep Unity so flexible. So sure, these goals can compliment each other (as we've seen in other engines) but this comes with it's own set of pros and cons too.

 

I'm glad you don't make commercial games, it's better to see Unity posting great stuff done by other developers rather than their own projects being pushed on the front pages, not that I've benefited from it, but I'm sure those who have loved being featured!

 

I think it would be cool if UT creates a flagship game which improves the branding recognition.

 

That is what they are hoping from us. To push unity and make something awe inspiring.

 

A single PBR doll or a small village isn't exactly pushing the boundaries is it? Even the UE4 tech demos, whilst orders of magnitude more intricate and complex aren't exactly pushing the boundaries either.

I think what some of us allude to is if you were to follow in CD Projekt Red's path for example, how far would you get?

 

Actually, the doll itself is made up of several parts and its purpose is to demonstrate how PBR in Unity can be used for a variety of textures and surfaces (stone, wood, leather, silk, metal, gold, ceramics etc) and how the U5 scene render settings can demonstrate the change in tone of different materials in a scene, on runtime with little scripting. It achieves its purpose incredibly well.

The viking village is to show how an environment can be created and how all of the new visual stuff (PBR, Reflection probes, Global Illumination, Scene Render Settings etc) fit in together to make game assets visually impressive. Also, the majority of the 3d models have been authored in a modular sense (For example: each house is made up of smaller parts, such as walls, doors, panels, cloth, roofing etc) so the village can be shrunk, extended or customised depending on what the users need are; the scene that is used for demonstration has probably 20ish houses/buildings.

 

I wish they would create an in-house AAA game even for once, so they can feel what we feel, how really simple low-priority bugs or features are important for us, stuff like SSR and good AA, make games way better, for the last few years we had problems with : PhysX, UI, x32 only editor, good post processing image effects, now some of them are being fixed, but if an AAA company were to choose an engine [over their internal], I think they would go with another engine and not unity, because simply:

when people search for Unreal engine they find this:


but when they search for Unity [5.0] engine they find this:


then they go deeper and find that games like Batman: series were made in Unreal, and they keep on searching for unity games and all they find are some platformer games, mobile games, they search deeper and may find The Forest graphically matching.

I do like unity a lot and it's still my number 1 engine, and I do agree that unity is easier to use, supports more platforms, but more platforms at the cost of what? late updates [semi-patch releases are fixing this]? exp. no OpenGL 4? no proper Input handling? headache giving terrain? no GPU particles? skin shaders? cinematic toolset? and sometimes feature x is not built-in, lets spend 3 months and build it, after the 3 months, unity makes the feature built-in, and there you go, 3 months of work wasted because they didn't go transparent about what they wanted to do [getting way better than before] and don't tell me you can always do it with the right team, when you want to choose an engine, you want to choose the one that provides most out of the box,

I'm not saying that unreal is better or vice-versa,
and someone may want to continue to post the pros of unity.

 

I do think that Unity probably needs to make more example games and not more example scenes. This should suffice for an extended learn team.

Over on UE4 (which is relevant here in discussion), epic staff comment that they had to change or add features to UE4 due to fortnight development exposing these weaknesses or missing features. So that hands-on first party feedback is essential. We all know that. Anyone kidding themselves is harming their own business model.

As it is currently too expensive and too far out of reach for Unity to boot up an inhouse game dev shop, the learn team offers the best next-bet - but only if they can increase the complexity of the examples, and game examples. I am sure they will once the majority of beginner stuff is out of the way. That is my hope for Unity, at least.

Hopefully the proper structure for a tight relationship between development and Learn team is in place. This vital inhouse feedback will depend on it.

Well, that's actually a) dumb and b) flat out wrong. It's irritating because a smart person shouldn't be saying things like that. It implies the user base is thick and easily manipulated by marketing. I don't think Unity wants to be reminded of this comment, so I think we should forgive them for it.

Unity aren't gods and nor are they our enemies. I consider Unity to be our partner - we must help them to help us. It's a relationship that is mutually beneficial.

I wish Unity would think sometimes before saying something that can be taken entirely out of context. If Unity continues to develop the Learn team and push out increasingly more complex game examples then this will not become an issue.

I feel the Learn team pushing forward with game examples of greater complexity (intermediate and advanced), will be enough to internally benefit from, teach, and externally impress.

 

Rather than Unity creating it in house, it would probably be preferable to work with or contract an outside studio to make something like UE4's Elemental demo.

 

Well, if they tried making a game they'd release an emergency update a month after and an apology for not having realized before how bad was the current terrain system.

 

Or cheaper & faster option:
Several competitions per year for "pushing unity visuals to the max", with bonus prize for making something really amazing (and if none of them are outstanding, they would get some regular prizes instead, like good enough sum of asset store credits to attract real talent)

Also its more likely that competition demos would end up faster in the asset store (for sale, but still available for anyone to learn), compared to 3rd party contract projects..

 

I think the reason they don't make games is because they don't want to deal with their GUI system.
After using 4.6 beta, I see now why everyone dislikes the old one.

 

With these two images all I can think is one of these probably runs on a three year old macbook alright, while the other is whipping a pair of titans in SLI that are crying in agony. I could kind of understand wanting a vertical slice of a game available to have a better benchmark for performance than a pure visual demo.

 

Can't we have both?

 

@andytouch

A single PBR doll as in one, a doll in single context even if it's made up of multiple parts. If you want a nice shiny doll you could do that with marmoset although I understand it's a learning demo (not pushing Unity as an engine exactly though is it?). How many houses do you think this has? The city is 8.5KM x 8.5KM, there are no "background" meshes here either. The thought of something like that strikes fear into an engine or game developers heart, we obviously can see the shear amount of awesome artwork gone into this.

Trying to create a game with all your rendering tech on this size or scale whilst getting it to run it on current generation hardware, now! That's an impressive feat of engineering skill and tech..

You can't begin to know or feel the pain of trying to achieve something half as impressive as this, until you do it and you realise how important your tools are to you. This is obviously an extreme example, neither does it need to be anywhere near this scale or technically magnificent in both the artwork and rendering side. But alas this is where things are heading, imagine what Enlighten would do if it looked at this scene. Something along the lines of ARRRRRRRRRRRRRRRRRRRRRRRRRRRRR!. As Hippo says, Epic are learning from games and as far as a 3D engine goes it's pretty well thought out. Even if it does like to cripple mediocre GPU's..

 

That's over exaggerated. I can run the Elemental Demo inside the Editor on my laptop (nVIDIA 570M), pretty well, no hiccups or slowdowns.

 

He was talking about that elemental demo.

 

I think Tato was, there is only one Elemental demo and it runs 60+ FPS on a 780m..

 

The one on the image is the original one with Voxel Cone Tracing which required 3 680s

I was just pointing out how even Epic had to discard some technology so it can run on inferior hardware in the same way Unity does to support mobile.
This is how it looks now:

 

I'm actually for Unity to create a game division, instead of plenty useless services divisions (UCloud, USocial, UName it). To my knowledge it's plain useless because there's tons of similar hosting, socials APIs and tools for Unity with a head-on and plenty expertise.
Having a gaming division (even if it's for mobile gaming) will put Unity in the front and add more value to it's main product the "Unity Engine". By creating games, Unity also targets more than game developers, it's targeting gamers all around the world. It could become a third (and very big) source of revenue for Unity.

 

Yeah, that's a slight change in particle count. I might actually get that to run on my poor old card.

 

Wrong,
That's the PS4 version, the PC version looks like this:


Just in case you want to try the PC version:
http://www.techpowerup.com/downloads/2368/unreal-engine-4-elemental-tech-demo/

 

Wait....What? You want Unity to start competing with us, bearing in mind they have the engine source and therefore intimate knowledge of how the engine works at its core(a clear advantage). Personally I think they should focus there efforts on providing, free quality materials, lots and lots of shader presets for said materials, and more tutorial style projects.

 

Sorry, this makes no sense.

 

Sorry this makes no sense.

 

Except it is Epic Games making actual games and not crappy little tech demos that allows them to fully understand how to improve upon their engine. Competition exists regardless of whether or not Unity makes their own games.

 

Clearly, a kid have left the building.

 

UE4 Elements Demo runs at literally 1 frame per second on a Surface Pro 2...that's pretty bad and unrealistic as to what you could do in a real game release...and personally I think the Butterfly Effect Demo kicks ass.

 

@Ryiah completely agree, as long as Epic are learning and improving the tools they provide us = more success! There is a serious dry spell in many genre markets at the moment. I'd welcome some more innovation!..

@Jonny Roy It's a PC and Console demo, it's not made for a tablet! Obviously it's going to run like crap. I'd like to see what Crysis 3 maxed out on that tablet runs like..

@tatoforever Quit with the insults, you're supposed to be a professional game developer.

 

As an Indie, I would not only be comfortable with but would also welcome UT developing and maintaining a game or two. I am not scared of Unity eating my lunch ( )

However, UT does sell their engine by the tons to professional game companies so I can see the problem there

Perhaps, one flagship game in each of the broad genres wouldn't harm Unity at all. Could even school the rest of us as to right way to get it done. A la Google and Nexus

 

I don't think it would be super advantageous for UT to make games. The reason behind them making games in this case would be because of the intention to better the engine, but how many types of games can actually be made with Unity? Isn't the whole point that the tools be so flexible you can do just about anything with them and do it on any platform?

That makes it a little silly in perspective since no matter what game type they decided to develop at any given time it would never provide enough practical and universally applicable feedback to the engine for the endeavor to be beneficial.

Better is to ask the existing community of developers that already aware of the engine's particular weaknesses in every type of genre and platform. The only downside is the barrage of negativity because of the perceived weaknesses, rather than the focus on exceptional flexibility.

 

The big difference is that any developer has access to the Unreal engine source code. I have come across a few 'undocumented' Unity features, one was for accessing the scene camera in the editor programatically The only reason the unity staff member said he knew of a way to do it, was because he had source code access http://forum.unity3d.com/threads/moving-scene-view-camera-from-editor-script.64920/ . A small thing, yes, but who knows what other goodies are hidden away.

Instead of that how about they focus on responding the user requests, complaints and bugs, as well as getting their engine to the standard of UE4 in terms of features.

 

@LaneFox

It's never anything to do with the "type of game" I see this thrown around so much, give me Visual studio or even QT and I can make "any type of game" using the OpenGL framework. But you need the experience and hands on to have a framework that supports it effectively.

Epic seem to be pre-empting features, e.g a TOD system can't effectively be lightmapped with baked shadows, so your stuck with CSM. Cool CSM that looks half decent is expensive , so you look into RTLD Shadows.

There we go, instant 20 - 40 % increase in performance, neither can you use baked reflections which is the only option in Unity 5.0.. So why wouldn't you start with a real-time reflections system with the option to bake for other platforms? I'll not even start with PhysX etc.

If the feedback system was efficient, Unity would be pre-empting the need. Corporate will evaluate the application and scope out limitations before use, so I'm not sure how effective the feedback model is. I've seen plenty of on the ball feedback from Indies and hobbyists alike, but I'm not sure how seriously it's being taken.

Some things you just have to try yourself.

 

Surely making a scene like that of the Witcher 3 and witnessing its horrible performance and then having to find ways to keep its performance acceptable without modifying the asset complexity or reducing its count would never provide enough practical and universally applicable feedback to the engine for the endeavor to be beneficial. Neither the terrain engine they would have to improve, or the proper high-performance vegetation shaders they'd have to come with.

You talk as if they can't do both. The staff who dedicates itself to creating complex scenes and game worlds has no need to be the same as the main engine development branch. Just like Crytek.

However, a Surface Pro 2 is a tablet, not a computer.

 

It's a pretty powerful one (as powerful as any laptop), which runs most modern games without any issues. If you are only targeting machines way above a Surface Pro 2 spec, you're probably talking about a very small audience.

 

How is an Intel HD 4400 GPU, powerful in any way shape or form? It's less powerful than a 7 year old mid range desktop card.

As powerful as any laptop you say? I'd like to see the benchmark scores between that and my Nvidia 780M gaming laptop.

 

Why not just ask the existing developers already trying to do that? Much more economical.

 

Why would it be more efficient to develop a game than do what they're currently doing in regards to this?

 

There are some things that can be only figured out when the staff is directly hands on with the project. And considering the state of features like Mecanim and the terrain engine, it is safe to say they won't take feedback from other studios as serious as the one they took let's say from the Butterfly effect made in-house.

Making games just gives you more perspective about where to take your engine.
Indeed: the only reason right now they know "where to take their engine" is because other studios have developed an engine. And those engines are all developed based on a game or games.
The important R&D part (knowing what is possible and can be done to improve games) has basically already been made for them. It is now a matter of to find out how to implement it.
But what if Unity had the best engine? Would they know where to take it next without no games or watching the competition?

 

Well it runs most games fine, but anyway this is turning into UE4 vs Unity, and there is plenty of threads on that. The reality is based on everyone referencing games that will only run on the top of the range gear should Unity internally try and create a game that can compete with Crysis 4 (As that's the next release) given that 1) It would mean dedicate a team of up to 1000 on a game for a few years, who could be working on making the engine better. 2) They don't have a team already creating games on a day to day basis so it would take even longer.

I think I'd rather them work on the engine so we can get some new features, if they had spent the last few years investing in games we might not be seeing the great stuff in U5 we're seeing now. Also, compared to the demos above, the Viking Village was used for the screenshot, what about the Race which looks great and runs on an Android Tablet, or the Butter Fly Effect which looks awesome even for an older demo.

 

just bring us more improvements, how-ever you do it, just do it, don't sit around for years to fix stuff like GUI, terrain, input, seriously, i think its just not right to have an R&D team of the size of nearly ~100 and actual stuff gets done by 10 people, it's ridiculous, I hope one day I see a unity internal developer and argued him into making any AAA game with unity [visuals], some stuff are way old to be in major updates, x64 editor? its nearly 2015 god damn it, PhysX 3? i think it was introduced in 2011, if they tried to make a game with unity, they would realize that they can't use the editor properly because it was x32, and next step they try to do their GUI, and well, we all know what, and why not try to create a keybinding options, or lets just support console controllers, smooth loading screens, good Post-Processing effects, after all they would realize that it's not happening, they can't create a game with their own engine,
I dare you to show me one game, only one game that can match current-gen graphics made with unity, look where games are headed, inFamous, The Last Of Us, FarCry 4, games are matching cinematic movies these days, we can't sit around developing for the Intel HD, if people want to play, they will have GTX 600+ or something powerful so don't start the low end argument, I don't want them to create games like Crysis 3+, I just want to create a demo that I can show people and they say wow, but actually be able to use those features, anyone remember all the cool stuff in The Butterfly Effect that never made it into Unity 4.x? some that won't make it into 5.x? I'm so done.