Are game engines the future of rendering?

Become a member of the CGSociety

Connect, Share, and Learn with our Large Growing CG Art Community. It's Free!

THREAD CLOSED
 
Thread Tools Search this Thread Display Modes
Old 03 March 2013   #61
And I have to add, I really really could use an Iclone like software that is easy to import objects to and follow normal 3d software paradigm.

(when iclone first appear, while cheap, require you to own 3d max to create character/clothes for it - which destroy the case completely).

There is no quad view in iclone either. (for a while). i can't remember now.

Good thing Source Filmaker is available now. But how friendly it is with Blender I have no idea.

Simson Movie tells that simple rendering is good enough if your story is strong.
 
Old 03 March 2013   #62
Originally Posted by teruchan: As nice as that would be, I simply don't agree with that. There are tons of films and TV shows coming out every week that don't look anywhere near as good as the best stuff in the cinema.

By that argument, there should be tons of film & TV companies making programmes using game engines then. Real world is not that simple though.

Sorting out the problems that a game engine pipeline throws up is no small task. For example getting assets out of CryEngine is a pain. Relatively easy to import but I believe can only export individual objects in .obj/.mtl. I believe all the extra stuff you can put in the sandbox such as voxel landscapes, procedural trees, decals, animation, reflection maps etc. can't be exported or are very difficult and tedious to.

Plus the 20% Crytek take of the money you make. I'm not sure what Epic Unreal takes or Unity for commercial projects?

Other questions like, how easy is the supporting software that uses the game engine? How well does it integrate with an existing pipeline? Has it been tested enough in the field? Cost for training and support and who does it? If it's just a one or two man act then it be difficult to provide all this in a robust manner.

By the time this is all sorted out, traditional offline ray tracing packages will have all gone GPU. GPUs will be that much quicker anyway and the growing pains of going down the deferred rasterising game engine route won't be worth it. Really big projects can afford to use render farms and the cloud anyway. So "are game engines the future of rendering?", I don't think so.

Jules
 
Old 03 March 2013   #63
game engine might be used to help the boost render time and create quick real time preview ( useful for mocap/bluescreen session) but not yet to finalize high quality render and post production.

also
the tool that used in game is hard to tweak , not really straightforward and barely physical/raytrace based.
the light system, is limited, is not vray where you can casually put hundred of light in one view , and expecting accurate photon bounce that mimic real world without artifacts.

- artist demand linear pipeline straight from 3d packages, not export to somewhere and build-it modular based workflow . no game engine integrated to 3d packages yet ...

SSAO SSself shadowing is no no for realistic light representation ... :3

but i still optimistic on how future realtime preview program can overcome this obstacle ..
__________________
http://arifcreations.com/

Last edited by Wongedan : 03 March 2013 at 01:35 PM.
 
Old 03 March 2013   #65
Originally Posted by Jules123: I think the best way Sebastian to gauge whether or not it is really working is if you get interest from studios wanting to invest in your process. Or maybe you make a great film (short or long that stands up to what's in the cinema) and people go wow how did you do that and how I can use that for my film?


yeah, that's what I'm trying to do. Also I like tinkering with the cryengine 2 internals and to develop new tools for rendering effects. It's fun, but I don't have much time and that's why it's a lengthy process.
It's not my life's purpose to prove and demonstrate cryengine viability, but I'm doing that anyway, so when I'll have something done with it, I'll post here for others to judge the technical aspects of it.

And I'm not saying everyone should switch to a game engine, but for certain situations, it could be useful even for final rendering. Even I will probably use also some 3ds max renderings along with real camera filmed footage and composite with the cryengine.

Originally Posted by Wongedan: game engine might be used to help the boost render time and create quick real time preview ( useful for mocap/bluescreen session) but not yet to finalize high quality render and post production.


Can you take a look at this post earlier in the thread and tell if you're still maintaining your opinion ?
 
Old 03 March 2013   #66
If you want blur cinema quality (raytracing) you can get very far with rastering techniques by faking: indirect shadows (ambient occlusion), indirect lighting (ambient light), reflections (reflection maps), glossy reflections (blurred reflection maps), reflection occlusion (masking reflections) and much more. Particles, motion blur, depth of field and color grading takes care of the rest. This is how they did it in vfx for many years and those thechniques are now implemented in realtime engines.

Perfect example is the new Unreal tech demo.
__________________
Follow me on Twitter
 
Old 03 March 2013   #67
regarding to Unreal 4 latest demo it still uses

From geforce site it says:

- Thousands of dynamic lights with tiled deferred shading

it might produce more consistent GI light resolution but doesnt remove the baking process entirely ... it something you still don't want to experience in film production.

still it promised to support thousands of independent deferred light simultaneously which much much better improvement !

as a game artist, baking (and exporting) is something I wish I never experience ...

Quote: Can you take a look at this post earlier in the thread and tell if you're still maintaining your opinion ?

yeah I saw that earlier , the cry engine jungle project looks cool! i want to see it when it finish.

but the CE demo/tool still doesn't convince me that would replace the film production pipeline yet.
Quote: game engine might be used to help the boost render time and create quick real time preview ( useful for mocap/bluescreen session) but not yet to finalize high quality render and post production.
__________________
http://arifcreations.com/

Last edited by Wongedan : 03 March 2013 at 02:11 PM.
 
Old 03 March 2013   #68
Originally Posted by Wongedan: but the CE demo/tool still doesn't convince me that would replace the film production pipeline yet.


Why would it need to replace the entire film production pipeline?

The thinking in a lot of this thread is that it's either one or the other, that'll you'll only ever use one renderer and that's it. That neither can exist in the same toolbox.

While at this stage I wouldn't render hero elements with a game engine, I'd certainly consider it for background elements such as crowds or environments.

Not every situation requires a 100,000+ polygon model, 100s of 4k texture maps, raytraced reflections and indirect diffuse lighting when it could use a 1000+ polygon model, 1x 1k texture map, mapped reflections and ambient occlusion lighting.

It also helps to think outside of what game-engines currently do - when dealing with film images your dealing with a different set of goals, so the game-engine when used as a game-engine is of little use in a film production environment.

To address some of the issues...

In film we don't need things working at 24-60fps, 1fps-1fpm is fine if it contains motion-blur, is anti-aliased, is written out as an exr and can do secondary outputs (position, normals, depth, diffuse, specular, etc). If it can't do secondary outputs in the same pass as the beauty then no problem, just need a shader which does it in a separate pass.

I'm also happy to bake things out, either once per job or once per frame or a mixture of both. These type of things when well-developed should become transparent to the lighting artist so that they don't need to worry about things.

The big thing holding us back from doing this now are maintaining hardware shaders (CGFX, HLSL, etc) which line up with software shaders (RSL, C++).

Obviously hardware shaders are feature-limited compared to their software cousins and to achieve the same look different techniques have to be used - the primary example of this would be subsurface scattering. And even if we ignore that aspect of it and deal with hardware shaders as simpler shading models, they still need to react to light in a consistent fashion with hero shaders.

At the moment there is no real solution to do either of these things automatically and it requires too much developer resources to do it manually as they either need to know how to write RSL and CGFX shaders or have a developer who maintains the CGFX shader library.

A year or so ago subdivision surfaces and displacement would have also been an issue to a hardware render, but things like OpenSubDiv and dynamic tessellation mean that hardware rendered geometry can be rendered in exactly the same method as software rendered geometry.
 
Old 04 April 2013   #69
Originally Posted by Jules123: By that argument, there should be tons of film & TV companies making programmes using game engines then. Real world is not that simple though.


The thread title is, "Are game engines the future of rendering?" Not the now. The issues you speak about with exporting would be moot points because they don't intend for you to be exporting anything. Just like with a game, the engine is the last step.

As with most things, it will probably start at the bottom. They already showed me one indie film being done with live actors on green screen and the rest by CryEngine. It looked awesome!

Also consider that, as in the Daz and Poser world, you can expect a massive content market to spring up around this, where soon there will be any type of asset you need, or multiples are varying levels of quality, to quickly be up and running on something you want to make. A lot of it will probably be free. Those who have the "ILM quality or not at all" attitude will probably never get into it. There will be tons of people, though, who will get into it, just like there are already thousands of machinima films out there, and maybe many people will call these films "crap", but people will be getting their projects done and out there.
__________________
Terrence Walker
Studio ArtFX
Learn How to Make Your Own Animated Projects!
You don't need millions of dollars or major studio backing!!
 
Old 04 April 2013   #70
Originally Posted by teruchan: The thread title is, "Are game engines the future of rendering?" Not the now. The issues you speak about with exporting would be moot points because they don't intend for you to be exporting anything. Just like with a game, the engine is the last step.

CryEngine been out for like 6 years? If it were so good for filmmaking, there be something out there on TV and cinema, not just YouTube surely? Isn't future correlated to the past when the quality of the engine was that good when it first came out.
Not a moot point for me, I couldn't even get depth maps out of the engine, it wouldn't let me :(

I think there are currently too many limitations and it will all rear its ugly head when you use it in production for real. You can if you have to spend so much time dancing around them, but I say why bother then?

The game engine wouldn't be the last stage for me, but its output used in the compositing stage. Did I say I couldn't get depth maps? So more manual fiddling than necessary.

Jules

P.S. I couldn't imagine doing a fight scene in a CryEngine virtual forest. You'd have to make sure there is no virtual foliage for the actors to step on, to stop them seeming to float above the ground, or passing through trees like ghosts. So many limitations, it's my belief that software around game engines need to evolve a lot more to make them really useful for filming with.
 
Old 04 April 2013   #71
Originally Posted by Jules123: CryEngine been out for like 6 years? If it were so good for filmmaking, there be something out there on TV and cinema, not just YouTube surely? Isn't future correlated to the past when the quality of the engine was that good when it first came out.
Not a moot point for me, I couldn't even get depth maps out of the engine, it wouldn't let me :(


Good point. I was thinking of exporting geometry, assets and the like. For those that wish to work that way, exporting passes for compositing might be a show stopping requirement. If me, I simply wouldn't do it. The film they showed me clearly didn't. It was basic green screen characters in realtime CG environments.

Originally Posted by Jules123: I think there are currently too many limitations and it will all rear its ugly head when you use it in production for real. You can if you have to spend so much time dancing around them, but I say why bother then?

The game engine wouldn't be the last stage for me, but its output used in the compositing stage. Did I say I couldn't get depth maps? So more manual fiddling than necessary.

Jules


I think it all depends on how one wants to work. If I was going into a project using such an engine, I wouldn't try to cram it into an existing pipeline requiring the engine to do what my current tools can do (export a dozen passes for diffuse,spec, reflection etc. etc.) I would use it for what it is. I would do everything in camera on the engine side and only composite the live action elements or other things as needed. I would approach it with a completely different mindset, or, otherwise, I agree as you say, "Why bother?"

Originally Posted by Jules123: P.S. I couldn't imagine doing a fight scene in a CryEngine virtual forest. You'd have to make sure there is no virtual foliage for the actors to step on, to stop them seeming to float above the ground, or passing through trees like ghosts. So many limitations, it's my belief that software around game engines need to evolve a lot more to make them really useful for filming with.


In that case, I would have a tiny area with real grass and plants for the characters to fight in and let the engine handle the larger scene around that. I don't think it would break the bank, assuming I am doing a super low budget indie film, to go buy a patch of turf and a few bushes. If enough scenes in the film required being in the forest, having some things for the actors to interact with would be a must I think.

I am not sure the forest or desert are best examples, though, as even low budget films still go off and shoot in the real forest or desert. I am thinking more of the sci-fi stuff, or things that can't be done in the real world.
__________________
Terrence Walker
Studio ArtFX
Learn How to Make Your Own Animated Projects!
You don't need millions of dollars or major studio backing!!
 
Old 04 April 2013   #72
A lot of actor performances are weakened when motion capture is performed.
There's a few guys like Andy Serkis who seem unhindered by the process, but there's always room for improvement.

In the next couple of years, with the introduction of technologies like Google Glass and Oculus Rift, I don't think it far fetched to assume that actors will have an opportunity to be immersed in real time virtual environments as they perform. In this regard, with these sorts of developments coming our way, I can see a lot of use for not just real time rendering, but gaming environments being loaded into headsets.
__________________
۩PRIST

 
Old 04 April 2013   #73
Where do you draw the line between a game engine and a renderer anyway? When FurryBall uses all gaming tricks to draw film frames and games use selective ray tracing, the lines are blurring. The two are only going to converge more as hardware gets faster.
 
Old 04 April 2013   #74
Originally Posted by stew: Where do you draw the line between a game engine and a renderer anyway? When FurryBall uses all gaming tricks to draw film frames and games use selective ray tracing, the lines are blurring. The two are only going to converge more as hardware gets faster.


Can we render games with renderman or Vray? Therefore those are not game engines. The line is that those are not easily interchangable. I'd love using those benefits of modern game engines, but I imagine the learning curve, and royalties are high. I'd love a tighter integration of let's say Vray and CE4, that we could use it as a plugin for certain rendering tasks.
 
Old 04 April 2013   #75
Originally Posted by Jules123: The game engine wouldn't be the last stage for me, but its output used in the compositing stage. Did I say I couldn't get depth maps?


Originally Posted by Jules123: P.S. I couldn't imagine doing a fight scene in a CryEngine virtual forest. You'd have to make sure there is no virtual foliage for the actors to step on, to stop them seeming to float above the ground, or passing through trees like ghosts.


Uh, so much naysaying , not only from you, but from many others. Of course you can get depth-maps, take a look on my cryengine website in my signature. And about interactions - you can get them as well, but of course you have to take time to prepare your assets just like in any program.
I will post again this gif I posted in a similar discussion here in 2010


Link with longer movie - 2.4 MB

And to answer something said by someone before - you can have high poly objects if you want - bigger than 100 000 poly. And even bigger with some workarounds. And also larger textures.

I'm planing for 2 years, I think, to post some technological demo videos about cryengine for film instead of games, but I didn't found the time. I hope I'll manage to make those videos in a few months, but rest assured you can do much more than you think.
And I'm talking about physics interaction as well, not only light and graphics. For example the plants can be tweaked to stay deformed to the ground after a tank passes by, and not spring back like they are default set.

Smoke and fog can also have interactions with wind, helicopters, light, shadows and so on.
 
Thread Closed share thread



Posting Rules
You may not post new threads
You may not post replies
You may not post attachments
You may not edit your posts

vB code is On
Smilies are On
[IMG] code is On
HTML code is Off
CGSociety
Society of Digital Artists
www.cgsociety.org

Powered by vBulletin
Copyright 2000 - 2006,
Jelsoft Enterprises Ltd.
Minimize Ads
Forum Jump
Miscellaneous

All times are GMT. The time now is 03:10 AM.


Powered by vBulletin
Copyright ©2000 - 2017, Jelsoft Enterprises Ltd.