Can Unity match CryEngine look while matching its performance?

Hi.
I am just interested in knowing: has someone been able to attain a CryEngine look map/level while maintaining the same level of performance? Or is it at least or will be possible with Unity 5?
Basically, right now I am only using Unity for 2D, but one of these days I'll be giving 3D a try and I want to know if I could make a level like the Crysis 1 jungle, without losing performance due to the high amount of ongoing vegetation, physics and destruction. But doing research, I have read some old Unity users abandoning ambitious projects because "the engine got to a limit that while we had a good look, the engine couldn't keep up". I have also heard that there's no way to match the speed of the build in Unity vegetation optimizations through scripting due to some low level optimizations being requires.

I am also interested to know if Unity 5 Enlighten will provide a lighting solution identical to that of CryEngine's Dynamic GI or if it is something totally different.

Do not even know why I ask because I am not moving to neither Unreal Engine [annoying sheen that hurt my eyes] or CryEngine [many much more talented than me have never released anything on it even after 7 years of usage to the extent it is called "the portfolio engine"]. Might as well make my own engine [will only take a few decades :-D ]
All that, and me liking Unity scripting.

And maybe not OT, but [I don't have the staff for this so it is just curiosity] think Unity 5 will be able to match Battlefield 4 graphics and performance out of the box by simple scripting, or will that still require getting a source license for doing low level optimizations?

In case someone asks what's the Cry look, that it is about big staff or art direction, I am refering to the countless vegetation single-user mod/showcase videos in Youtube like this:

 

For doing something like that, a lot of it just comes down to whether you have the talent/resources/manpower to make it. For example, in terms of post processing effects that video has HDR bloom, vignetting, lens flares, light shafts, and motion blur. Unity has all of those (although you'd have to get a third party motion blur on the store since the built-in motion blur isn't too great). All you need after that is mostly occlusion culling, a good LoD system and high quality art assets.

Unity has Umbra occlusion culling built in, and while I haven't used the Pro LOD system, rolling your own is straightforward enough. So things aren't too tough on the programming side. So the main thing you'd be worrying about would be making the 3d assets along with their LoD meshes. Obviously the whole thing would look much better with Unity 5's GI system, but other than that I don't see much reason why you couldn't do something like in that video in Unity. There's proof it can be done to some extent (as far as handling that volume of assets anyway), and unlike that video, this one includes actual gameplay on top of rendering a very pretty forest.

 

In short: you would have a hard time matching Crysis 1 (a 7 year old game). Don't even try with Crysis 3.

Maybe you can match the look, but performance would suffer badly. And read all the papers about all the effects in Crysis. You cannot just buy some assets from asset store. Maybe it would not look bad, but not the same. For Crysis they invented a whole lot of very subtle new rendering techniques, the professionals eye will immediately spot that it is only "almost Crysis". Redoing those in a way the performance is still good is an art only very few programmers can achieve. You better try to clone flappy bird than Crysis.

 

Short answer: No.

Long answer: No. Unity's performance for a lot of its built-in image effects even is so poor that you may as well not even use them. The amount of time you'll spend just getting performant fullscreen effects out of Unity is actually a little mind-boggling. To add to this, GPU particles only work on a single platform and you'll be writing your own code for those as well. The best you can hope for is a reasonable approximation through faking most of the effects.

 

Crysis has a ton of new rendering effects they invented. For a quick and dirty clone you are maybe right, but those rendering effects matter, it is maybe subtle so people not reading about this don't see it, but it is what makes the mood.

If you look at the details of those 2 videos you see that those two things are not the same. The fog/bright light in the Unity one is not for mood, it is a trick to not kill performance. But both videos do not show the real test cases, like walking down a straight road from a hill top in clear weather (with another forest hill top in view), or something like that. Like in Crysis 1 where you can see over the island. Doing that is art. Just rendering trees around the player and making sure he cannot see very far is not the same.

Don't get me wrong. What they do with "The Forest" is very impressive. But comparing it with Crysis 3 makes no sense.

 

The crytek engine is multi-threaded and version 3 can use up to 8 cores. Unity cannot, it does offload some work to another core but the engine only runs on one core. Run your game or a demo and check the windows task managers CPU %, then run a crytek 3 game and check again.

 

Wait until Unity 5.0 comes out and then see if it's possible or not.. I'm heavy into CryTeks technology, I love the way they do things but don't underestimate the power of enlighten coupled with speedtree for terrains..

 

I guess it is all down to performance.
Think there's a chance they implement those rendering techniques with Unity 5?
No way to implement them with scripting, right? [not that I have the ability, but I have heard some things are better done at a low level]

It looks really good but yeah, like someone said the fog effect feels more like if it was being used for better performance rather than direction due to how close some vegetation LoDs are set, like the ones at 1:04. The fog fits the game mood and atmosphere, but the distraction due to LoD swaps gives that impression. Not that I am that nitpicking when playing a game, but I am talking about high amounts of vegetation at the distance without the engine flinching.

 

Easily the most realistic looking forest scene I've seen from any game engine. Wonder if it's hold up without the fog and with noon sun? I'd like to see if such a demo scene would even run on my PC.

 

Most of the post will be shader based (you can do fog that way too), the core lighting functions and middleware like Enlighten you can't touch. But why would you? It's better than an LPV solution (IMO). After lighting and post it's just more shaders, you have to understand that CryTeks ILLUM shader for example is amazing and their approach to tech as a whole is very well researched and thought out.

Although I like Unitys approach, can't beat them? Get in a company who can .

 

On a tangent; In my opinion making a forest look great in a video isn't the same as making a forest look great in a demo. I see a lot of this crap online where people post a video and say "look at my demo video" and I think "wow impressive". Then I download it and it looks only half as good. Like many model vegetation assets have great videos but when you download the actual product are crap. Don't be suckered in unless you have actually tried the real program.

 

That user generated forest is far above what AAA games have even on max detail levels. It's a portfolio piece. Even the game doesn't contain that detail levels. So you're not making tonnes of sense, no.

You can do that in Unity if you must, but there's little point unless it's just a tech demo. What matters is how well something performs under game stress.

 

ATM no one seems to have a reason to make something like that in Unity. I wonder if AAA quality assets, imaging effects and shaders would be popular enough on the asset store to pay themselves off.

 

It's not just that nobody has reason to, but that it's prohibitively difficult to do in a way that both looks good and is performant, which is a huge problem with the ongoing quest for photorealism in games.

 

If someone had a reason to do it they would overcome those issues, big cash money wads would be the key but ATM I'd imagine the Asset store wouldn't have the sales to suffice. A game perhaps but then it comes down to whether or not its worth extending Unity to do this or not.

 

If somebody had a reason to do it, they could just do it on an engine better suited to the performance needs of that sort of endeavor, since the cost of extending Unity to do that is well beyond trivial.

 

No, it can't, not without some hard work, but why? IMO, CE3 is a tech demo engine. I have yet to see a good game made with it, and it's graphics are "blurry". I like clear graphics, where I can tell a baddie from the scenery.

 

Maybe the question should be how much would it cost to do this in Unity?

The problem being that you can probably get the CryEngine Toolset and out of the box throw a forest together with this level of detail and scale. With Unity you cannot, so how much extra will it cost in additional assets / work hours to achieve the same results?

 

That is not the question at all, we have a 24KM2 world machine tiled output with highly detailed terrain foliage. We have built our own LPV solution and all terrain is async'd and we have our own post.. Unity at the moment can not handle it, you have to use dynamic shadows because whilst importing / exporting that amount of detail to beast the editor will fall over plus the performance is horrid, prefab streaming is hit and miss to say the least and the SC performance is rubbish and looks ugly.

The engine has to be built from the ground up to start managing this sort of stuff, whilst the look is mainly lighting and post which isn't too difficult. Level streaming / multi-threading / 64-bit editor is a must, there is no way around it.

In CE you just slap some trees in, tweak some settings and it runs fine.. That 24KM2 terrain full of foliage ran @ 100 FPS on a 6850 and that's including the 500,000+ poly ocean they have going on..

I will be testing out Unity 5.0 to see how capable that is of doing large detailed terrains.

 

LOL, I'm not making sense?

1) I don't own a console.
2) I don't play AAA games so I couldn't tell you that this in fact exceeds all of them, simply that it does for those that I have seen.
3) The graphics of AAA are vastly overrated as is the game play. I just read a game review for the first time ever of the new Wolfenstein (being as it's the 1st and one of the few games I've paid for, it's a bit sentimental so I decided what they are doing with the IP 22 years later) and graphics wise it looks as if it was made with assets I've bought in the asset store. Really.

further...

4) I simply state that forest scene made in a video posted to this forum and generated by a game engine is the best I've ever seen.
5) I wonder, knowing something about my PC whether or not I could download that project and create a video that looks as good on my PC.
6) Other threads are absolutely howling that UE4 and CryTek shame Unity in this regard because they want such a nice realistic look in their games but I'm told in this thread it doesn't make sense that I would want do that myself.

7) Yet the selling point of new consoles and new versions of game engines is the graphics but my interest in such a beautifully done forest doesn't make sense?

MS is buying game IP as a selling point for xBox but MS Sony mostly depends on improving graphics otherwise the Wolf3D I was given 22 years ago is as good or probably better as the newest released version of Wolfenstein. At least if you're to judge by the bizarre comments left by readers at the review web site (Kontacku or something like that). Really.

Why?

Is walking through a forest too boring? Perhaps I want to create a botanist training program?

So unless you create cruddy graphics and the uninspired game play of the second video in this thread it's just not worth doing? Must everything produced by these game engines conform to some type of horror / crime genre with inferior graphics to be 'worth while'.

Maybe this Speedtree can create the needed assets for botany training simulations? Eh? In Unity 5? That's not worth doing? And here I thought civilized folk appreciate being able to differentiate between eating poison and eating food. Or in my case watching the skin of my right hand peel off like a glove after pulling what I thought was English Ivy off the side of my house. (It was poison ivy)

 

I'm not sure what goat is going on about ..

But anyway, the example provided is far from the best I've seen come out of CryEngine..

 

Whatever. :mrgreen:

 

goat, I'm talking to the op not you. I'll be sure to quote him next time so you don't rise in fury and start butting my immovable hippo arse

 

I'm getting sick of threads like this. I think people are confused because Unity doesn't have that many graphically impressive AAA quality examples. Most of the prominent games made in Unity are either indie games that are loved more for their unique gameplay than their AAA visuals or mobile games that look really good for mobile games but don't look AAA console / pc quality.

But, you must ask yourself when looking at a video like this: does Unity have all of the features that make this look the way it does? For the most part, 99% of the time the answer is yes. The next question is if YOU can make visuals with that level of quality. If not, then you are the only thing preventing that level of quality , not the engine. It might be debatable that it is easier or harder to do X thing in X engine, but it can't be debated that the "big 3" game engines all share most of the same features at this point.

Once again the "matching it's performance" part is entirely dependent on how well you can optimize. Unity gives you all the options you'd ever want to optimize visuals, it's up to you to use them properly. Sure there might be small differences, but at the end of the day a small performance drop or boost should not be the sole determining factor of whether or not you should use an engine. Do you like the engine, is it easy for you to use, etc. are all way better determining factors then "this engine can run 1% faster under the same circumstances as this other engine".

I think a lot of people think that posts like this are empty air and they need to see examples. That's fine. Here's a video of an asset pack that you can download yourself in Unity and see for yourself exactly everything that is done to make this video. I think you will agree that this video looks very similar both in aesthetics and quality to The Last of Us, which was released last year (not that old) and was generally praised very highly for it's graphics.

 

Absolutely nothing but camera pans happen in that video and even then it seems that it's struggling to get out of the 20s when it comes to FPS.

 

I'm not noticing the performance drops that you are, but even if they do exist you are making a lot of assumptions. You have no idea what the video card is that rendered this video, and recording software vastly increases overhead.

Once again, you can feel free to get this asset pack yourself and run it on your computer. I guarantee you it will run fine assuming you can run modern games.

 

LOL, well I already know to keep my mouth shut so no excuse. I tend to read unquoted threads chronologically. I'd still like to create a decent looking virtual botany forest in Unity; well maybe I'd better start with a garden.

 

Have you every attempted this rather than just saying whatever you believe and getting sick of other people mentioning this stuff?

Can I personally make something like this? Yes I can, not only do I have pretty much every Manu4K pack out there I also have a decent sized art department that can improve on these visuals, with access to the likes of the Quixel Megascans / NDO / DDO and a PBR workflow.

Can I make post effects, you're damn right I can!.

Can I improve on the shaders? Still you're damn right I can!

Can I build a 64-Bit editor for Unity and introduce level streaming! HELL NO! Simple as that, you don't have access to low level engine code, YOU cannot create current "Western AAA's" in Unity 4.0! . Could I make a lighting system as good as Enlighten? NO, not a chance in hell! Could I even match CryTeks implementation of LPV with there 20+ R&D tech engineers? NO! Even to do an integration with middle ware type SVOGI I'd need to plug in.

Can Unity 5.X cycle do all this AAA junk? Why yes, yes it can!

Let's compare: CryEngine:

 

You don't have to look twice to see it's done with unity (shivering&shaking bushes and trees, all grass bending at same time to same direction, static sky...etc.etc). Assets and art style are good.

 

@ChaosWWW:

That scene looks awesome, but...

1) One does not see the details of the assets. E.g. do all assets have detail diffuse and detail normal map in addition to the standard ones? Is the engine able to trigger them without the user noticing?

2) Having some fixed blocks or a whole huge level is not the same. And the camera in the video moves in a way that is very nice to the engine.

3) From what I have read just having the post processing effects a standard AAA title has nowadays makes Unity too slow even without many assets. And having many active entities seems to kill performance too (I follow a guy who wanted to have 1500 units on screen in a RTS, he gave up using Unity for that).

 

Looks like Kudzu factory but it was better example of Unity then the 2nd post in this thread.

There were obvious billboarded plants. I wouldn't have know it was Unity in particular if I hadn't seen those assets in the asset store though but the one thing I didn't really, really like was the stiff grass blowing in synchronicity. Unity would have did a better job optimizing then those billboarded plants but you have to allow that Unity or any other game engine isn't going to give great performance with a game environment setup to try and look like you're flying on a magic carpet at the height of the Amazon tree line with a focused vision to a depth of infinity.

 

(1). Yes, I am a game artist and I primarily have used Unity, but I have also used Unreal Engine 3 and 4.

(2). As you point out, all this stuff is coming with Unity 5.

(3). Looking at this video (I have seen it before), from the looks of it none of this stuff couldn't be done in Unity 5. Unity 4 can still achieve graphics that you find in most games, since most games coming out now are still targeting last gen consoles, and thus don't have pbr usually and definitely don't have GI (it's debatable if GI can even be done effectively on the current consoles). This demo looks better then most games you can actually play. If you want your final game to look better then a tech demo then, well, you have really high standards.

In general I get what you're saying, I'm really looking forward to Unity 5 as well. But Unity 4 doesn't prohibit you from making a good looking game. I'd say only 5% of the games that came out within the past 2 years even have these awesome features.

1. The trees look fine, they look like they have a light breeze going through them. You can tune Unity's wind to have a higher frequency in stronger bursts is you want more intense wind.

2. This is the only good point in this list.

3. You can make a dynamic sky in Unity.

1. It doesn't matter if these particular assets do or not. You can have detail maps in Unity if you want them, and I don't even know what you mean by the second question. They would mip-map out like any other texture. If you wanted to you could make the LODs have a different shader without the detail maps.

2. This is an asset pack. If you wanted to make a game with this pack without the fixed camera, you could, and it would still run fine. If you don't believe me you can feel free to download it and test it for yourself. I can tell you I've bought one of Michael O's asset packs and I have had none of the issues you seem to be thinking exist. Once again you don't have to take my word, if you are actually curious about this feel free to run your own tests.

The whole "camera is very nice to the engine statement" is nonsense that is demonstrably false. You're going to have to elaborate on that.

3. Okay, it's nice that you've read that, but there's nothing stopping you from running these tests yourself. I think you'll find Unity's post effects do no under perform within your own projects. And having multiple entities is naturally going to bog down any engine, but there are ways to optimize that as well.

 

He means you can make your game look better by dividing and conquer the environment, not just the game play, in smaller levels and not to put everything in one level like you want to create the Russian Federation.

 

I know you see it as your job to defend Unity from every perceived slight, but this is absolutely ridiculous. Unity's post-processing performance with the included effects is absolutely abhorrent and rolling your own is incredibly time consuming and not something that other engines have to deal with to nearly the same degree. Unity has severe advanced graphics performance issues out of the box and getting it up to speed is a time-consuming and complicated process.

 

It's nice that you are passionate about this, but using extreme words doesn't make it true. Can you demonstrate this with any evidence at all? Your argument seems to imply "it's well known that Unity's post effects are bad", but this just doesn't hold up. In my experience this is far from true. You are making the assertion, so you have to provide some sort of argument outside of acting indignant.

As for myself seeing "it as your job to defend Unity from every perceived slight," I do like Unity and as such do find myself defending it, but only against complaints that are false. Unity has many problems that people bring up that I don't defend because they are legitimate problems.

 

Add some more intense wind and branches start to disconnect from main branch, that's why you have to do this kind of shaking and shivering

Can you point me where i can enable dynamic sky i'm bored with this static stuff.

 

Just add some volumetric clouds and and a small camera on a slow rotation then have your second camera only render that w/ overlay (Layered effect). It's not that hard to do..

For some reason some artists believe tech isn't important, even though the modelling apps they use have some of the most advanced lighting and rendering functions on the planet. If that were the case we'd be stuck in the ID 1 era..

How did you find coding all the performance tweaks in Unity?

How many times did the editor bottle it? You generally have to end up scaling back heavily or putting months into your own LOD / OC systems to get the best out of it. Without the bolt ons / tools, it's not an engine it's a renderer..

Exactly, but as it stands right now.. You can not do in Unity what you can do in CryEngine.

Don't disagree in some respects, In others I do.. What AAA games has included good looking global illumination I wonder? CryEngine and ones with Enlighten someone says? Well I never.!

So I'd double check before you want a healthy debate, not to say that baking isn't in some instance a lot better anyway. But we should know for a fact that GI methods are perfectly valid.

PBR agree on that and some AAA games look pretty poor, I've seen better come out of Unity. But still when we are talking CryEngine here it's a different matter.. There are reasons why you don't see many games in it though, not related to this thread.

Multiple issues here with Unity 4.0 and it mainly stems around the editor, a big amount of meshes and terrain sent to beast will bowl it over. To do effective OC and LOD, you have to do it yourself. All the post really does need replacing if you need decent performance and a good look, I'm still trying to decide what there AA shader does because it doesn't seem to get rid of jaggies that well. Too much terrain, or meshes = Unity falling over, select too many large images = Unity falls over. Out of the box you REALLY need to replace all the shaders, or hope to hell someone on the asset store knows what they're doing. Dynamic shadows suck, best to bake (if you can) and so does dynamic lighting which again Unity 5.0 will sort out and so will the inclusion of a full DR.

In no way is Unity 4.0 a western AAA engine.

When you're making AAA games (or even A / AA), you're not aiming for what a game looks like now. You R&D to see what it will look like in three years, we are developers and the latest tech now will be defacto at the time of your release. That being said, I have got our rather large game running on some pretty bad hardware!. I think our big cave scene now runs with 250 draw calls and 6.5ms render / MT time.. All out you're looking at 500 draw calls and 10 ms. There are thousands of meshes in this scene, runs like a dream.. Not to say CE still isn't better at it, because it is!.

I truly believe Unity 5.0 will be a turning point and all this silliness will come to an end.!

 

Oh another engine zealot thread...

Seriously, who cares if Unity can match CryEngine, or UE4 in look and performance?

99% of all people here are not be able to produce the art content for such games anyways, let a lone a full and polished 3d game. So why bother?

 

"hello, I'm nipoco and I'm defeatist as all hell."

 

It's not the size of the engine, it's what you do with it.

-- ShadowK (c) 2014

 

I know what i can and can't do with Unity, but doing all these tasks by yourself... that's not the goal is it. This was about Unity vs CryEngine tools and quality. How does this volumentric clouds "dynamic sky" compare to this http://docs.cryengine.com/display/SDKDOC2/Setting+Up+Time+Of+Day ?

 

1. Honestly this hasn't really been an issue for me. With some extensions and passable scripting knowledge I've been able to get what I've wanted unless I had absolutely no clue how to implement it (I.E: SSS would be nice but I have no effing idea how I'd do that..) For those latter cases, I agree that the engine should support me. In general I haven't had to deal with large scale projects up until now, but for larger scale projects (like the one I'm doing now), there should be at least one programmer / tech guy who can solve the performance issues and create new systems. I can't speak for these people and how difficult / annoying their job is in Unity versus other engines.

2. This is true. For me I really want these features, but am unwilling to switch from Unity due to how much I like it on other fronts. I'd rather wait for Unity to catch up, especially since these features are confirmed, it's just a matter of playing the waiting game.

3. I think I worded this one a little badly. I'm saying, for one, that there is some debate about whether or not the current gen consoles are going to handle real time global illumination without sacrifices in other categories. It's definitely possible to do, as engines like Cryengine prove, but it's viability is harder to determine. For example, Crysis 3 on PC has global illumination but absolutely did not on the console releases (which were last gen). It's still early this current gen, so we'll see if any console games have global illumination, but as far as I know none have it so far (I could be mistaken. Does Ryse have GI on the consoles?).

As for your other points, I basically agree with you. I think we're basically on the same page.

 

@ Imre keep your socks on , I'm working on a TOD and weather / water system as we speak. Point being like you said it's a pain in the ass, I'm having to do a lot of work for what comes in CE and UE4 already. But my point is, it's possible even if it is a PITA. Would I like this included? Hell yeah, if little old me can do it than Unity guys defo can.!

@ Chaos, if I'm coming off a little harsh I apologise. But I too get tired of the evangelical engine thing, truth be told all engines have flaws and they all have what the other one needs. As things moving on they are all merging into the same thing, Unity will be a AAA worthy engine in it's next incarnation (As long as they don't mess it up).

If Unreal was back in 2010 before they did there upgrades we could say the exact same thing about them too. I think Crysis 3 on PS3 and Xbox was just a shed load of lights to imitate GI. PC had bounces.. But BF3 did use enlighten, because it's damn good!

 

There will never be the one perfect engine. Only engines that fits your needs, or don't.

But what do I know. I'm just a defeatist :mrgreen:

 

Well it wasn't me that said that ..

 

I know Shadow.

Was not aimed at you

 

These types of threads are kind of pointless, but weighing in:

-If you had the choice to make a game in CryEngine that had "10/10" visuals, or a game in Unity that had "9/10" visuals (eg. The Forest), you should use Unity to make that game. End of story. To make the same game in CryEngine you'll probably spend twice as long (if not, longer).

CryEngine is a tech demo engine, and not much more. Maybe, if your whole team had only ever worked in CryEngine before and never worked in Unity, then you would choose it, but apart from that, there really is not much point.
If anything, CryEngine's visual fidelity scares people away from buying games - take a look at recent stats from Unity that show amount of CPU cores/RAM/etc for standalone player PCs - you'll see why being able to use 8 cores is not so important.

 

But now you're talking about something else here!! Using it, not capabilities. It looks pretty, but there is not a chance in hell I'd use it to make an actual game with.

 

This thread was done with this in mind:

Basically, it would be a waste not to use Unity when it is much faster to achieve things than CryEngine and I am already a bit more familiarized with its scripting. I was wondering if it was better to leave this engine as my 2D tool while learning another for 3D so I came to ask for the general consensus.

I had seen that Manufactura K4 asset video, but for some reason his videos are lacking some "feeling of depth" present in many of CryEngine textures... which made me also wonder if even the way different kinds of mappings are hard coded into the engine affect their perception. [Different algorithms make a difference in this? Can someone clarify this? I also compared it to RTP, they don't give me the same perception]

But seeing I am just starting with 3D [or rather, the texturing 3D part which is the one that gives me trouble] if Unity gets there in 3 years, I won't complain. Though that makes me an egoist because some already want [and have wanted for a long time] that level of quality, and some even go as far as saying "boy, my assets become AAA when I move them to that engine!". I was propelled to make this thread after seeing one of the best Unity screenshots in one of ShadowK's thread, and then reading the engine had trouble keeping up.

And "not enough staff to produce art content" is never an argument. You don't know if I am willing to take 3 years to make a nice and polished 512m^2 city :-D

 

Yeah maybe not. Especially if you make that just out of pure enjoyment/hobby.

Indeed you can make a triple A looking game as single dude, if you're dedicated and skilled enough. It just will take a very long time, which is, from a commercial standpoint, not desirable in most cases.

And Unity5 + some add-ons will be almost as much capable, as UE4 and CryEngine3.