This topic is locked from further discussion.
Of course we aren't. Have a look around the room you're sitting in. Go have a look out your window. If realism isn't your thing then sit down and watch Wall-E, Up, Toy Story 3, Avatar, etc...
As soon as the average game starts looking like the world around you then this argument can be made.
We are getting close to a point of diminishing results with models. Not saying that we are anywhere near it honestly, but that is one aspect that has been pushed much further than others and with dynamic tessellation for LOD we are making leaps in that direction.
But lighting, character animation, shaders, physics, particle systems, volumetrics, etc.. are all still very much in their infancy when it comes to video games.
Oh and not to be a drag, but SSS shaders (sub surface scattering) has been around for a while and has been used in countless projects. No need to tag it to LotR, or Avatar like its something new. If you've ever seen a convincing render of skin, milk, marble, wax, etc.. you've seen SSS and there are already game engines that support, or in some cases emulate such shaders very well.
This is all very CPU demanding, and tells me the future of gameing may be in having a s**t load of cores, and ALOT of ram, I.E. in the 10's of gigabites (it wont even be perfected until we are into the HUNDREDS of gigabites of ram, which will prolly be the gen AFTER next). I think people should be expecting alot more ram from next gen consoles (at least Sony's console, and maybe even nintendo). keep in mind this is all still just rendering still scenes without anything going on, even with 10's to hundreds of gigs of ram (the polygons are all drawn by the CPU and the only constraint to this is ram). And even then, they only get about 5 fps just to move the camera around. Not only is ram what devs are, we should hope for a console with about 16-32gbs of ram. I just made this thread for people who think current PC graphics are anywhere close to hitting "realism". movies like Avatar? Those movies take DAYS for computers with a sh*t ton of ram to render. They render the movies slowly at a few frames per second and let it sit for days, and then speed it up to the proper framerate. We can barely render real time VIDEO rendering of ray tracing without it being a slideshow. Videogame graphics have a LONG way to go before we hit Avatar realism. StrongDeadliftWhen we get to it in games most of the raytracing will be done on the GPU. Current dx10 and dx11 GPUs can do decent raytracing already, not for any serious game in realtime, but alot faster than any desktop CPU. Look up Luxrender or octane render. CUDA and OpenCL renderers that work on your GPU. Also there is the Nvidia design garage with is their demo of raytracing for newer cards.
And another thing in the current generation of graphics we are going into the realm of real time GI with limited bouncing light. Crysis 2 on PC has the light from the sun bounce, i believe 5 bounces. Things like geometrics enlighten(BF3) and other real time GI systems can have bouncing indirect light as well. So not raytracing, but a step closer
Next-generation games will be primarily using tesselation, not ray tracing; it provides an impressive step up (see Epic's Unreal Engine 3 Samaritan Demo) without the enormous performance hit ray tracing currently carries.
Why is Ray Tracing always considered the "end all" aspect to graphics? I mean I've read about it, understand it and it looks great, but arent there other less taxing solutions that look just as good?
I read somewhere that Pixars Renderman doesnt even use ray tracing. And isnt radiosity taking over as a solution for both games and film? I am a noob so I'm asking this seriously.
Next-generation games will be primarily using tesselation, not ray tracing; it provides an impressive step up (see Epic's Unreal Engine 3 Samaritan Demo) without the enormous performance hit ray tracing currently carries.
garrett_daniels
Aren't those unrelated though? Tessellation from what I understand subdivides and displaces polygons as the model gets closer to the camera whereas ray tracing is a very accurate way of generating light and shadow.
So I think, again I don't know for certain.
They aren't things you compare. Games don't use tessellation "instead" of raytracing, they are in completely different catagories. Tessellation isn't a method of lighting, it is something that increases mesh density.Next-generation games will be primarily using tesselation, not ray tracing; it provides an impressive step up (see Epic's Unreal Engine 3 Samaritan Demo) without the enormous performance hit ray tracing currently carries.
garrett_daniels
[QUOTE="garrett_daniels"]They aren't things you compare. Games don't use tessellation "instead" of raytracing, they are in completely different catagories. Tessellation isn't a method of lighting, it is something that increases mesh density.Next-generation games will be primarily using tesselation, not ray tracing; it provides an impressive step up (see Epic's Unreal Engine 3 Samaritan Demo) without the enormous performance hit ray tracing currently carries.
ferret-gamer
That's what I was thinking. Sorry to be a bother, but you seem to know about these things. Could you answer my previous question? :)
I see I wasn't very clear at all. Sorry about that. :)
Tesselation and ray tracing work in completely unrelated ways but are used to achieve the same result, i.e. significantly better graphics; the difference here is that ray tracing carries a huge performance hit, so we'll be seeing games using heavy tesselation with little or no ray tracing--especially since existing lighting methods already look pretty good but it's the models and textures that aren't quite up to the same standard.
[QUOTE="garrett_daniels"]
Next-generation games will be primarily using tesselation, not ray tracing; it provides an impressive step up (see Epic's Unreal Engine 3 Samaritan Demo) without the enormous performance hit ray tracing currently carries.
ConanTheStoner
Aren't those unrelated though? Tessellation from what I understand subdivides and displaces polygons as the model gets closer to the camera whereas ray tracing is a very accurate way of generating light and shadow.
So I think, again I don't know for certain.
You're right.
The two are compltely unrelated.
And by the time hardware gets powerful enough to do real time ray tracing, we'll be talking about photon mapping :)
Realistically, we're looking at rasterization methods witha ton of improvements for the next 10-15 years.
The thing is, you can get 99% of the way to ray-tracing with rasterization. It's just a matter of improving certain ways of rendering light/shadow/meshes, and throwing more GPU power in to the mix.
In other words, you can do the stuff that is being done win those ray-tracing videos with rasterization, and you ca do it with less hardware muscle. That will always be true. So the quesiton becomes, why do ray-tracing?
I guess the answer will be a rhetorical: why not? But that won't happen until we're running MUCH more powerful ahrdware.... Or to be more accurate, when the server farms that will be streaming our games into our smart TV's and phones, have the hardware capable of doing so.
I see I wasn't very clear at all. Sorry about that. :)
Tesselation and ray tracing work in completely unrelated ways but are used to achieve the same result, i.e. significantly better graphics; the difference here is that ray tracing carries a huge performance hit, so we'll be seeing games using heavy tesselation with little or no ray tracing--especially since existing lighting methods already look pretty good but it's the models and textures that aren't quite up to the same standard.
garrett_daniels
Ray tracing = lighting/shadowing solution.
Tessellation = 3D displacement mapping which inrcreases the detail of 3d models.
Not really related at all.
[QUOTE="ConanTheStoner"]
[QUOTE="garrett_daniels"]
Next-generation games will be primarily using tesselation, not ray tracing; it provides an impressive step up (see Epic's Unreal Engine 3 Samaritan Demo) without the enormous performance hit ray tracing currently carries.
Kinthalis
Aren't those unrelated though? Tessellation from what I understand subdivides and displaces polygons as the model gets closer to the camera whereas ray tracing is a very accurate way of generating light and shadow.
So I think, again I don't know for certain.
You're right.
The two are compltely unrelated.
And by the time hardware gets powerful enough to do real time ray tracing, we'll be talking about photon mapping :)
Realistically, we're looking at rasterization methods witha ton of improvements for the next 10-15 years.
The thing is, you can get 99% of the way to ray-tracing with rasterization. It's just a matter of improving certain ways of rendering light/shadow/meshes, and throwing more GPU power in to the mix.
In other words, you can do the stuff that is being done win those ray-tracing videos with rasterization, and you ca do it with less hardware muscle. That will always be true. So the quesiton becomes, why do ray-tracing?
I guess the answer will be a rhetorical: why not? But that won't happen until we're running MUCH more powerful ahrdware.... Or to be more accurate, when the server farms that will be streaming our games into our smart TV's and phones, have the hardware capable of doing so.
Hey thanks for the info!
*goes to read about photon mapping*
Pixar uses raytracing. I think what you are talking about is them having a very mild to non-existent GI effect because they want the artists to have maximum control over the lighting. Ray-tracing is considered the end all, because it is the best way to approximate real light. Light rays bounce in real life, and that is what ray-tracing does. Radosity is nice, and can be implemented to a limited degree in real time for video games, see Battlefield 3, and will be the direction video games will be going for a while before raytracing because it is possible to implement in real time for games. There are other possibilities of what we will see in advancement for lighting methods like photon mapping, or stuff like CE3 is doing, but more likely real time radosity will become prevalent next gen. In radosity vs raytracing, raytracing is much better at giving realistic lighting results, specifically for shadows, reflections, and refractions, and the only major limitation is performance.Why is Ray Tracing always considered the "end all" aspect to graphics? I mean I've read about it, understand it and it looks great, but arent there other less taxing solutions that look just as good?
I read somewhere that Pixars Renderman doesnt even use ray tracing. And isnt radiosity taking over as a solution for both games and film? I am a noob so I'm asking this seriously.
ConanTheStoner
[QUOTE="ConanTheStoner"]Pixar uses raytracing. I think what you are talking about is them having a very mild to non-existent GI effect because they want the artists to have maximum control over the lighting. Ray-tracing is considered the end all, because it is the best way to approximate real light. Light rays bounce in real life, and that is what ray-tracing does. Radosity is nice, and can be implemented to a limited degree in real time for video games, see Battlefield 3, and will be the direction video games will be going for a while before raytracing because it is possible to implement in real time for games. There are other possibilities of what we will see in advancement for lighting methods like photon mapping, or stuff like CE3 is doing, but more likely real time radosity will become prevalent next gen. In radosity vs raytracing, raytracing is much better at giving realistic lighting results, specifically for shadows, reflections, and refractions, and the only major limitation is performance.Why is Ray Tracing always considered the "end all" aspect to graphics? I mean I've read about it, understand it and it looks great, but arent there other less taxing solutions that look just as good?
I read somewhere that Pixars Renderman doesnt even use ray tracing. And isnt radiosity taking over as a solution for both games and film? I am a noob so I'm asking this seriously.
ferret-gamer
Awesome, thanks for the thorough answer. And there goes that photon mapping again! :P Must read more.
I'm interested in getting into digital art, not necessarily for games, but I'm finding this subject to be much more complex than I imagined.
Also for the hermits to play with, here is LuxMark, a raytracing openCL benchmark, it has an interactive mode where you can try to move it around while your GPU trys to render ithttp://www.luxrender.net/wiki/LuxMark
Mah score is 9785 :D
You mean Overrated 3: Drakes Deception :P. Seriously, i would quit gaming if that were the pinnacle of gaming graphics.But we don't need that because we have Uncharted 3! :P
o0HAPPY0o
If anything, I imagine what they most likely do is they use SSS to create the the massive production models, and then take the normal maps from them to create the in game model you play with.Oh and not to be a drag, but SSS shaders (sub surface scattering) has been around for a while and has been used in countless projects. No need to tag it to LotR, or Avatar like its something new. If you've ever seen a convincing render of skin, milk, marble, wax, etc.. you've seen SSS and there are already game engines that support, or in some cases emulate such shaders very well.
sandbox3d
One thing I am curious about, and keep in mind im a noob, is what is the relation to Ray Tracing and Deffered Rendering? I remember back before Killzone 2 (yes two) was coming out, people kept talking about how it used deffered rendering, and they always mentioned it in the same breath as ray tracing, like it was similar in concept or something. What is Deferred rendering and what is its relation to Ray tracing?
those PCs aren't struggling to produce those images AT ALL.BrunoBRSDo you notice how in the car video, when he moves the camera, all those artifacts show up? Also, they only get like 5fps, all of this is just rendering a STILL scene with absolutely no animation whatsoever. Making a PLAYABLE game out of it, never mind an actual GOOD game, nevermind again an AAA game, nevermind again a 60fps framerate, nevermind even a 30fps framerate. Also, at the end of the first vid, the guy shows the highest number of polygons he got on screen (which was 2.3 billion or something) and it took 260gb of ram to render, and it only got 5fps.
If anything, I imagine what they most likely do is they use SSS to create the the massive production models, and then take the normal maps from them to create the in game model you play with.[QUOTE="sandbox3d"]
Oh and not to be a drag, but SSS shaders (sub surface scattering) has been around for a while and has been used in countless projects. No need to tag it to LotR, or Avatar like its something new. If you've ever seen a convincing render of skin, milk, marble, wax, etc.. you've seen SSS and there are already game engines that support, or in some cases emulate such shaders very well.
StrongDeadlift
One thing I am curious about, and keep in mind im a noob, is what is the relation to Ray Tracing and Deffered Rendering? I remember back before Killzone 2 (yes two) was coming out, people kept talking about how it used deffered rendering, and they always mentioned it in the same breath as ray tracing, like it was similar in concept or something. What is Deferred rendering and what is its relation to Ray tracing?
The first part of your post make too much sense to me. Are you mixing up SSS with a sculpting program? Game devs have been creating very high poly models and then baking normal maps for awhile now, it is common place in game development.Deferred rendering is basically a method of rendering light sources that allows for a crapton of actual light sources, but it has restrictions and disadvantages compared to forward rendering, but with dx11 and other advances those disadvantages are lessening and deferred rendering is becoming more prominent. It isn't often mentioned in tandem with raytracing, actually i rarely hear them together. I know some devs(Metro 2033) use deferred rendering to create bouncing light for a few light sources because they have the capablity to have a few hundred or thousand light sources in a single frame, but it isn't actual light sources.
Kz2 was one of the earlier deferred rendering games, and toted it alot so it got hype because of that. Raytracing was mentioned a few times in regards to KZ2 because this was still the time where some idiots thought the ps3 was a super computer and having a helmet with a specular map meant raytracing.
Please Log In to post.
Log in to comment