Unity Cannot Produce Good Graphics....I Think Not

My vote is crap engine, as the character's shadows is represented as a huge blob I'd say it's not a current version of Unity or UDK..

I'd love to be surprised though.

 

Thanks for the feedback on the shadows. For that, this is a mobile game, not a desktop game. I'd be interested to know what shaders since adding shadows to mobile games slows it down significantly or not supported on certain Android devices. We've made lots of work-arounds for the shadow effect.


In mobile, lots of things have to be down-scaled for it to work. Heavy shaders are no-no and effects - SSAO, bloom would cause crash instead. DirectX things are not support as mobile uses OpenGL.

 

Sorry Joe , didn't know it was an up coming one from you guys..

Have a look at Joe Dever's Lone wolf, it's been made in Unity for mobile and it looks amazing. I mean it looks better than a current crop of desktop RPG's.. The man's done some major work in optimisation and knows his contrast ratio's..

http://www.youtube.com/watch?v=altVUFXkk-E

It's possible with a lot of work, but I don't get too heavy into mobile optimisation as mine's aimed at desktops..

 

You're welcome and thanks for the feedback.

 

Cheers for the info, poked me in the right direction for getting everything setup at home. Ive used max since 2000, but never in a professional capacity, perfectly aware of baking AO etc but not used max's shaders to work with a model. I should be just on principle able to move a shader between glsl, cg and hlsl so its a good reason to brush up on that (And make necessary adjustments between implementations). I don't really have any desire or need to work for anyone but it would be nice to have a workflow follow a similar path.

And no worries was aware of everything in the links too, my comments on tonemapping and colour changes etc related to how you presented your work on a showreel or a webdemo to a prospective employer, not anything inside max

 

Well you can have a laugh at mine for now, needs a lot of work.. But getting there slowly!.. Still got a lot in from the asset store but fading out one at a time.. Any feedback is welcome.!

 

Anytime, I'm sure I'll need plenty at some point myself

Have you done much with Vray?

 

@ ShadowK

Yes! Rather familiar with, and fond of vray, been using it quite a long time, i dont think that's necessarily proof that im any good with it and i tend to be a bit conservative with my tweaking the GI, i definitely lack the touch of people trained in its use in a professional context, but i guess i get some brute force success. It's generally my first choice for lightmapping or baking textures straight in, it can make things look ridiculously pretty, obviously you lose lots of the options given by beast but since i generally use it for mobile things its not so big a problem

I recently compiled a bunch of webplayers of unity stuff ive done over the last 2-3 years and if theres any baked lighting in the whatever it is it's almost certainly done with vray http://pourfoi.co.uk - It has mobile endeavours there too with controls redone a bit, apologies for scrappiness (i find the goblin falling through the floor funny, however)

I'm not sure if you intended to give pointers of such but that stuff kind of explains my understanding of it

 

Saying "$75 a month isn't that much, it's just as much as going out drinking with friends" relies on the conceit that everyone can afford to go out drinking with friends.

 

I showed it to the lead artist here. He recognised some of the shaders. They are, and probably guessing:

- NGUI free shaders. Humm. Didn't know you could use the shaders from that.
- Tashren Water shaders.
- The rest seems to be some ready-made shaders from the Stumpy Shader set, the rest are custom shaders from that toolkit.

 

I like it, cool stuff LazyGun .

I recon people with an insight to the likes of Vray knows exactly what a powerful rendering pipeline can do.. Sure a lot of it's not achievable realtime, but these post processing effects make simple geometry look so realistic it's scary.. I was consulting for an API in a rendering house (cinematics) and told the guy I was into games dev..

We sat down had a coffee and he showed me a simple scene with a simple class room geometry and ran it through Vray, I was stunned.. I'd never seen anything like it. Ever since then, I've never once thought it's all in the artists hands because some of them tools out there are nothing short of mind blowing. It takes a lot of skill, mindblowing amounts of skill to get it right but impressive doesn't cover it..

So as a technical artist using these tools, do you also think it's not all in the hands of an artist? I honestly think deep diving into both sides is an eye opener.

Like this, a Vray scene done in 3dsMax.. Which has a metric crap ton of post processing.

http://www.youtube.com/watch?v=qHD8Xf5Rnvo

 

Awesome stuff Joe, I hope it helps

You should also be able to increase texture resolution by selecting a different compression algorithm and get away with it, DXT1 might cut it?

It does make a huge difference from automatic compressed..

 

Not really, but it depends how you approach it.. The scene is 1KM X 1KM around 5000 trees and a shed load of grass, we wrote our own optimised shaders for a lot of it. Not only that we have our own developed water based on Water Pro, which is taking around 250 draw calls in total the whole scene including lighting and shadows takes up 2500 draw calls and 58 FPS on average on a radeon 6850 which is relatively old tech now with graphics maxed out.

Switching off DX11 and mimicking tessellation shaders nearly doubles the FPS, we have been experimenting with optimising AA techniques although they have relatively light impact anyway, the game comes to a halt in forward rendering so we had no choice but to use deferred lighting and HDR (Which IMO looks better anyway).

Hard shadows with a smoothing technique massively improves performance and the 2.5K draw calls is before we have done any light mapping, probes and occlusion culling. Finally the target is 1000 - 1500 draw calls @ 70 FPS when the scene is finished with graphics set to beautiful.

 

I'm imagining you're referring to vray rt, which is their gpu renderer and it's startling, its not currently considered 'production quality' but its ridiculously fast and when you're rendering stills over andover to pick up slight changes in a material its breath of fresh air. My main woe is it doesnt render to texture. Everthing ive just said may be rendered (hah) void with the release of vray 3 (asuming its not been relesed already, i really should pay more attention) as rt in 3 is intended to be a valid replacement for the cpu renderer, without a few options natually, but it was more stuff like it being able to render.. render elements which got me hoping render to texture as well because that would make my year

And regarding who has responsibility for the art, i think everyone does, and i think it's backwards not to at least err, encourage? learning how a shader works for example at whatever role you play on the art team. Of course i dont have experience of studio work, i had a great team earlier in the year who were just enthusiasts working for a competition, we were lucky that one of the modelers was the gf of a tools/art guy at guerilla and he was a great help and seemed to agree with me on a bunch of stuff including ideas like having a room with say 6 unity set machines no matter what engine the actual production is using because a room full of enthusiastic clever people being told to 'make cool stuff' for a game could very quickly create all sorts of results and it be an energetic natural process. Concept programming. Something that wouldnt have occurred to me prior to using Unity.

But yes maybe its not practical to coordinate that kind of education, teaching artists a full understanding of the process, and the more technical guys understand the considerations of the artists, but it does seem very backwards to me when surely you'd enjoy a vastly more efficient production when the tech folk and art folk can directly communicate ideas to each other using shared, understood language and concepts instead of considering each other aliens

So yes, i dont know how that translates to reality, as far as i recall Valve for eg hugely encourage their staff to be mobile multidisciplinarians, might not be so practical in a slightly less lofty setup, but assuming i was hired (not sure i can fulfil any specific role strongly enough and noone is going to employ a generalist with no serious industry experience), i would be very disappointed that a big bunch of my skills would stagnate, and I feel never be addressed.. not very enthusing a company to work for. And god forbid i was hiring, well, i'd not be placing myself in that situation if i was going to compromise, i'd employ people competent in their field and smart enough to develop a communication with any other discipline in the production. Easy for me to say hypothetically though i guess

 

OpenGL does not support DXT* compression.

 

Nice scenes, ShadowK but I have doubts it would run well even in Cryengine. This is something that becomes a real issue for fill rate (assuming there's decent batching).

One of the things I consider important when judging engine quality, is merely the base speed of how quickly it can render. This above all, is your bottom line for serious development as more will usually mean more art, more detail and so on. When it comes to certain things like grass, you have to leverage GPU features where possible, and this is less of an engine thing and more of a case study into unique techniques and shaders etc.

 

Sorry Joe, I'm not sure about Unity.. DXT1 is supported in OpenGL, I forgot it's not supported on mobile though it's PVRTC.. Or you could try just uncompressing them .

http://www.opengl.org/registry/specs/EXT/texture_compression_s3tc.txt

 

Thanks Hippo, the original plan was to build our own engine geared towards this type of game. Whilst the rendering pipeline and effects were impressive it was so unstable and couldn't be used. I'm talking GPU alert computer shutdown from overheating unstable, after a long time of trying to get the GPU dumping working correctly I eventually gave in.

Peformance wise, at the moment Unity is running faster than CE. We got 30FPS in a scene similar and a lot of crashes, so no doubt Unity is the way to go at the moment.. It's just things like the editor running slow (64-bit editor) and a couple of other features (CMAA, SSDO, LPV, RGI) that would push it that extra 5% which could make a world of difference, well to an OCD sod like me maybe anyway so when the games complete it's still competitive, if you get what I mean. You got to think, what's the base line of graphics going to be in three years? Where will the competition be at?

Unity is more than capable as is, but I'm excited to see where it develops in the future..

Thanks again and very much agree.

 

This might only be related to performance issues in Unity but: I read that CryEngine 4 (not yet released I believe) benefits from an amazing (and obvious) LOD solution. It creates each version with one click and stores them as needed. This provides infinite joy to the technical artists in mid-pro as theres no need to manually crunch the mesh, lower texture resolutions and simplify shaders for each level. Just one click.

Player is 10cm away = 2048 X texture maps, 4k V's and a shader to shame the VFX boys.
Player is 10m away = 518 X maps, 500 V's and bumped specular.
Player is 1000m away 52 X maps, 100 V's (or even cutout) and a diffuse shader.

Player doesn't notice a bloody thing all thanks to extreme ease of implementation.
But thats Cry Engine 4..... *sigh*

 

I'm pretty sure Unreal/UDK does this as well with its SimplyGon implementation.

 

We're targeting for the mobile platform. The scenes you saw only have max. 30 - 50 draw calls. The most on one terrain is 70 draw calls.

In some scenes, the draw calls are barely legal - below 18 at around 2,000 - 1,500 FPS. On the iPad 2 and Android devices, it registers at nearly 60 fps.

 

How does the employer afford to pay all the employees and equipment?


I hope you are not working for those non-existent companies or working for company with bad credit ratings.

 

You do realise that not everyone who uses Unity has employees, r-right?

 

Is Unity's version of Simplygon capable?

 

Haven't had a chance to use it, sad to say. I'm only just starting into 3D work with Unity.

 

Have you considered raising income by accepting small tasks, building a portfolio of works or go find a non-Unity job to support your hobby?

 

I bet Florian has no idea I'm re-posting all his Unity stuff here...

 

Have you considered not making assumptions about everyone else's situation?

 

Well then, good luck with your Unity project.

 

Conceit?
If you aren't earning more than $75 dollars a month and is not from the poorest third world country, never mind drinking or even game development, not the least the affordability of Unity, you got a way bigger problem than that.

 

The florian stuff seems to be:

- marmoset
- amplify motion blur
- that dude's battlefield screen smudge thing

So that's a lot of plugging stuff in that could be doubled in frame rate with some effort (probably uses a fair few redundant passes), but it looks fantastic.

 

Yep. In theory it should be easy to do. I'll try it when I get the chance and post the results. The song, for those who wondered, is "Can't kill us" by The Glitch Mob. It's a teaser for their new album which looks amazing. I recommend anyone who their music it to pre-order on iTunes.

 

Yeah, I totally don't need to pay a lease or for electricity or for the gas for my heater or for my internet or for water or for food or anything like that.

 

Looks pretty, but try and run that real time.. I quote from the man himself:

Some good tips in there.

 

I honestly don't care much for graphics. If the game is fun and playable that's all I care for!

 

Please spread the word, it'll make my life much easier.

 

It does run real time, in webplayer no less.

There are the limitations with anti-aliasing.... and such, which is what you quoted. But Florian stuff is like movie quality, you can't complain!

If you think you can't go past those limitations, just take a look at Da Bawss video, which provides an example of such quality on a larger scenery; http://www.youtube.com/watch?v=mub2lKel9JE

The question is if Unity can produce good graphics, the answer is yes.

It doesn't come 100% out of the box... but what does? Does all the work come out of the box? No, and thank god, otherwise everyone would be releasing the same gimmick. What's the point of that? What you want is achieving a wider amount of things with less effort, than achieving one thing with no effort at all.

Unity is just the tool for you to achieve what you need, with less effort.
That's all. If it's AAA quality, you've got it pal! Of course, you still need a good team, and likely a big budget as well, if you want to pull of a large scale AAA game. That's a given.

Easy big projects with next-gen graphics are not going to happen, at least not for a long time, because you still need to put some thought on optimization, quality, aesthetics, etc. The extra tech research you might have to do with Unity, compared to engines like UDK or CryEngine, is nothing compared to the work it takes to complete a full game. So what's the point on having "AO and Bloom" out of the box? Have you seen sloppy stuff on Bloom, AO and such? Not impressive at all right?

 

@ dogzer

That example is no where near up to par with current / future title AAA level games, impressive as it is Florian's example IMO is the closest I've ever seen compared to what CE gives you out the box and there's still extremely nervous geometry, bad anti aliasing and color reflection is no where near as impressive as the stills. Well I'm not one to look a gift horse in the mouth, it's encouraging to know at least indoor scenes can look kind of impressive.. (In still photo's)

"If it would be a complete game, there would be lots of things that would kill the quality and I wouldn't know how to avoid it."

There's the issue, in CE3 the effect would be less noticeable and the tech's been available a fair amount of time now . That's if you every got your game released!.

Look Unity can produce good graphics, we all know that.. Does that mean we shouldn't nudge Unity to keep progressing? No! Also there appears to be a heavy reliance on the asset store to fill features in here.!

It be interesting to hear any deficit's are due to the artwork with Florian's example, he's done a cracking job.. Hence the stills looking so good. Webplayer not so much!.

 

@Shadow

Yes. Totally, Unity Pro users should ask and expect Unity to improve the graphics quality. And I'm pretty sure they will, based on their impressive level of commitment so far.

Easy, out of the box, access to better graphics is not unwanted. Benefits of improved graphics in small indie games would be nice indeed.
It's just not super essential for small indie developers, unless you've got movie quality assets, and you have a hawken's art team, or something. But my point is, if you have such budget and art team, it also means you have the means to expand Unity.

Have Unity team made graphics kind of secondary, while other engines have made it a primary thing? Maybe, in comparison. And not considering Unreal Engine and Cry Engine have had tons of previous AAA experience.

Is that a bad thing? If you think about it, no. Because now Unity has a solid base, a great workflow for small indies.

While UDK is forced to re-write everything as they want to change their workflow... and next UDK does look very promising, but problem is... is it near the horizon?? I mean, we saw a yaw-dropping features walk-through more than a year ago, any news on that? Release date is N/A

For all I know, by the time UDK 4 is released... Unity (with DX11 already available) will have played the "top-notch out the box graphics" card long before.

 

How much is the source code for Unity?.. Believe or not UE4 with source isn't that expensive if you need it right now.

Whilst people with larger teams and budget's don't expect things to be handed to them on a plate, there's a fine grey line area where feature set's ease the burden of developing.

I agree, it's all about workflow.. It really is, some competitors border on unusable because of it. I'm personally not a fan of the GRAPHIX! mentality, but it seems everyone is focused on getting the one up on everyone else graphically in some sectors.

The only people who matter in this equation is the end user, the people who buy your games and it's common sense, don't sacrifice your game from graphics.. Keep competitive with the latest graphical features, they don't care how you did it and generally don't care what you did it with.. As long as the graphics are good / competitive and the game-play doesn't suck your on a winner.

Anything Unity can do to save us time is greatly appreciated, even if there's a rise in cost to do so. Compared to 3dsMax UT pro is chump change.

 

I wouldn't be surprised if there were some major graphical improvements to Unity coming in the next few major updates. I remember a long time ago there was a NinjaCamp demo for indirect illumination, for instance. Something like that, especially if cross-platform, could help a lot.

 

I find it the heap of these discussions popping up lately, a wee bit irritiating.
I've been seeing so many beautiful looking things made in Unity, yet we now have 2 threads talking about how it still isn't enough.

And it's only my personal theory...I would tend to believe that 80% of the peopla complaining about Unity not being up to par neither have the knowledge nor the available artists around to even get close to Unity's technical limits.
I never saw some technical nicely done game/asset around here in which the creator ended up saying "yeh well, nice but still Unity doesn't make it look beautiful enough - I go switch engines".

To me it always seems like it's more about the "UDK has more presets and assets, shaders to toy around with to get the same looking stuff" compared to Unity being more diverse in the core available stuff.
I don't know how UDK forums look like but I sure saw lots of things made in UDK that didn't really look pretty at all, "buttugly" would be flattering, in fact. Even if you have the most beautiful stock assets in the world an non-experienced user can make it look horrible in a heartbeat. I tried

I don't even know to which of these 2 threads I should reply to, since the same arguments are made in both by the same people, more often than not...They could be merged easily, or even better, closed...
Nobody is forced to use any of the available engines, neither is Unity forced to change it's roadmap due to some 20 users not liking it. As others have stated, for every user complaining there are at least 1000 not stating "I totally love what they are doing and offering"...
I find it quite astounding how much Unity has improved within this year. Just have a look at the showcase projects, wow, how did that mature and excel. It's just wonderful. And so far, Unity staff has taken to heart what bigger streams in user movement have wanted or wished for.

For me, personally Unity has evolved in a great direction given the latest 2D feature update, obviously. But I'm just one user and I totally understand others wanting different things.
However if it's only 3D gfxgasms, well then it was clearly stated there are already other engines out there that serve this appeal for now. Just take the leap and switch to these until Unity has improved on that. Nothing holding anyone back. I doubt anyone purchased a copy with the implicated right to have UDK like stock gfx and whatnot within a year of devtime...
I would think one chooses the tool with it's current available feature set in mind for a particular project and not which features might be included somewhere down the line.

I would have loved if there was at least one thread around christmas time, thanking the hardworking people over at Unity - just to see that it would never reach the 8 pages of non-ending discussions any other "Unity lacks this and that" threads.

I love my Unity

 

The problem is that getting to Unity's technical limits is a long and overly-involved process that can be done in other engines somewhat more effectively out of the box. This ends up equating to more work-hours put in for the same effect.

 

This is the way I see it

Unity:


UDK/CryEngine:

 

Hehe nice one dogzerx and yeh it is pretty much like that.

There is also other way arround. You can make bad looking assets look very good in good render-engine. Thats why i was saying year before that by default models do look better in UDK or CE , becouse there are already good candies loaded on default. Now luckly we have most of those candies on Asset store too, really making Asset Store for Unity was best freaking thing they have done since start of this engine, i said it before and i am going to say it again.

 

proof? I don't see any evidence of this. A bad asset with an ugly texture or bad modelling in relation to everything else, will stay ugly. You've got to have it in relation. So lets say it's 1 bad asset among a lot of lovely ones. That bad asset will stick out like a sore thumb. Or you've got all-bad assets but beautiful engine? looks like a pile of crap across the board.

I don't think this comment you made, will ever make any sense. A turd with depth of field, bloom and shadows is still a turd. You can't play that turd either. If you put it in a proper game, it will stick out like a turd. If you have a lot of turds, it's a lot of turds with lensflare and bloom, it's meaningless.

 

Do you know what the default UDK scene is?

It's a platform with a cube on it. Neither of these are textured. But you know what? That cube looks F***ing astounding.

 

We should clarify what's a "bad asset".

A white cube is a clean shape... it's not a bad geometry my any means, in fact, it's perfect!

It presents no obstacle to UDK eye candy.

Who wouldn't agree Minecraft becomes uber awesome with those DOF shaders things, provided anyone can run that stuff at an ok frame rate.

A simple character, without much detail, yet harmonic, and tidy, can go a long way if you place it on a scene with AO, DOF, Motion Blur, blah blah.

But try to fix a character with bad proportions, bad texturing and UV, sloppy geometry, etc. You can render it with Pixar's render engine if you want, it wont look any better!

 

For S***s and giggles, I decided to try and replicate the default UDK scene using nothing but Unity Free and all the default features. Here's what the scene looks like untouched, with just the cube, surface, and default lighting (aside from a directional light because otherwise nothing would be visible):

http://i.imgur.com/YlfDsIV.png

Wow, barf, what a turd! It's just a cube on a surface which might actually just be another cube I scaled awkwardly for the sake of the demo! So what happens when I do things like enable lightmaps and shadows and antialiasing and skyboxes and a whole bunch of other S***? Well, that cube turns into a beautiful princess.

http://i.imgur.com/pqlvgXR.png

You can make S***ty assets sing, you just need to work for them.

edit: That said, there's still a lot of problems. Unity's shadowcasting is absolutely terrible when it comes to colours. Like, shadows are pitch black aside from the ambient light, so you're left with all sorts of weird issues unless you very specifically adjust everything and carefully position lights. You could probably (maybe?) script something so that you always have a maximum light level and then dynamically adjust the ambient light, scene lights, and scene light balances, but that would be overly complicated and prone to wash out the scene.

I honestly can't wait until Shader Forge is out just so I can start faking some basic lighting tricks.

 

@Murgilod

It's all good with a cube... until you're trying to sell something.

That's when having actual polished art comes into play, not to mention the workflow that allows you to make it all work and look alive.