[quote]
because the performance required for DX11 is more than the performance required for DX10, as was DX10 to DX9. if they added dx11 features to dx10 in an update, people with 8xxx cards or 3xxx cards would just bitch about not being able to pull off this or that feature.[/QUOTE]
Thanks for clearing that up.
I'm going to address a few points.
[QUOTE=Odellus;20131356][IMG]http://farm1.static.flickr.com/186/437944396_b2f73a2996.jpg[/IMG]
thread[/QUOTE]
Concept images =/= actual implementations.
[QUOTE=gman003-main;20095752]a driver update could probably bring an old GT7200 into DX11 compatibility[/QUOTE]
Uh, no. This isn't something that's implemented solely in software, you need a DX11 card to take advantage of this, as the tessellation is done in hardware. So unless you know of a way to digitally update the actual GPU core, then you're going to have to get new hardware.
[QUOTE=johanz;20122721]then why not just model a high poly mesh and put it ingame?[/QUOTE]
As stated by multiple people, there's a reason we haven't done this. It's not just memory, the gpu's essentially storing the tesselated vertices anyway, and most of your VRAM contains textures and render targets and stuff. The problem is bandwidth/CPU power. If you've ever worked with a graphics API, you'll know that you don't upload a model to the GPU and then tell it to draw. You send each face to the GPU. Meaning the CPU has to process that and then send it over the northbridge to your GPU. If we had models with billions of polys, then the bandwidth and CPU usage would quickly become bottlenecks. Putting the tesselator in hardware and then implementing displacement mapping in a vertex shader with texture fetch is a much more space and performance efficient method for adding detail like this.
[QUOTE=luishi5000;20105798]I think we are over-looking some other aspects of DX11 that are, in my opinion, more important. First of all, lets talk shadows. <shadow image>
[/QUOTE]
I hope you realize that all that's doing is altering the sampling kernel based on the difference between the pixel depth and the projected depth shadowmap. Something we've been able to do for a while now (perfectly possible in DX9). And there's loads of SSAO demos out there for DX9. It seems like developers are trying to push new versions of DirectX by holding back features for older versions. The only other reason I can see for this is that improvements in other areas of the API have left more budget for effects like this. (See M_B's post above)
[QUOTE=BMCHa;20132653]I'm going to address a few points.
Concept images =/= actual implementations.
Uh, no. This isn't something that's implemented solely in software, you need a DX11 card to take advantage of this, as the tessellation is done in hardware. So unless you know of a way to digitally update the actual GPU core, then you're going to have to get new hardware.
As stated by multiple people, there's a reason we haven't done this. It's not just memory, the gpu's essentially storing the tesselated vertices anyway, and most of your VRAM contains textures and render targets and stuff. The problem is bandwidth/CPU power. If you've ever worked with a graphics API, you'll know that you don't upload a model to the GPU and then tell it to draw. You send each face to the GPU. Meaning the CPU has to process that and then send it over the northbridge to your GPU. If we had models with billions of polys, then the bandwidth and CPU usage would quickly become bottlenecks. Putting the tesselator in hardware and then implementing displacement mapping in a vertex shader with texture fetch is a much more space and performance efficient method for adding detail like this.
I hope you realize that all that's doing is altering the sampling kernel based on the difference between the pixel depth and the projected depth shadowmap. Something we've been able to do for a while now (perfectly possible in DX9). And there's loads of SSAO demos out there for DX9. It seems like developers are trying to push new versions of DirectX by holding back features for older versions. The only other reason I can see for this is that improvements in other areas of the API have left more budget for effects like this. (See M_B's post above)[/QUOTE]
It was a joke.
[QUOTE=PelPix123;20123551]Nope. Shadow blur is caused by the area of the light and the way the photons are emitted from that spread of light. The blur you are seeing is effectively separate shadows from an infinite amount of mathematical points emitting photons over the area of the light.
If the light was a point with no area, the shadows would not be blurred.[/QUOTE]
Who the fuck cares about the technical shit.
In real life, if you walk outside, shadows become blurrier as they extend farther outward. Therefore, we would like to replicate that in games.
[QUOTE=SomeRandomGuy18;20134319]Who the fuck cares about the technical shit.
In real life, if you walk outside, shadows become blurrier as they extend farther outward. Therefore, we would like to replicate that in games.[/QUOTE]
it depends entirely upon where the light source is in relation to the object
[QUOTE=PelPix123;20134059]This video pisses me off though. It sounds like he truly thinks that DX11 supermagically makes everything detailed without any extra work, and it's just the magic of running any old game through DX11 [/QUOTE]
This. Sad thing is, 90% of people are like him. Too many people believe these retards who hype shit up, but don't even know what they're talking about. And in the end they all go crying about how DX10/11 is useless because the graphics in dx10 versions of their games aren't all that different from dx9. Would you really expect the developers to completely rewrite their graphics engines just for those few percent of people who have new graphics cards?
[QUOTE=redwinterwol;20094821]im ganna wait for a second gen fermi card :P
also DX11 just looks like it would make the ground look painful to walk on[/QUOTE]
See you in 2012. :D
[editline]04:21PM[/editline]
[QUOTE=janky;20095180]Pretty cool. It looks like they used normal or parallax maps to actually detail the model rather than just creating the illusion of doing it.[/QUOTE]
They just use simple displacement mapping.
[editline]04:32PM[/editline]
[QUOTE=PelPix123;20123551]Nope. Shadow blur is caused by the area of the light and the way the photons are emitted from that spread of light. The blur you are seeing is effectively separate shadows from an infinite amount of mathematical points emitting photons over the area of the light.
If the light was a point with no area, the shadows would not be blurred.[/QUOTE]
Are you dumb or playing dumb to troll? Don't disagree if you don't disagree. There's no such thing as a shadow that is sharper in the distance than at the point of contact. Point sources do not exist in the real world.
parralax occulsion mapping in clear sky:
Note that it doesnt go past the corner because the polygon ends there.(crysis has the occulsion mapping inwards so it can do that.
[img]http://www.tweakguides.com/images/ClearSky_11a.jpg[/img]
AT this point the thread has become a debate on science/pseudoscience as well as a debate on the vast/minute differences between Dx10 and Dx11.
You're all dumb for busting a nut over this shit.
This a shitty demo as well. Biased, just like all past DirectX demos.
But does tessellated geometry get a modified physics box? If not, it will be like in crysis, floating or doing ugly collisions.
[QUOTE=Doug52392;20108626]Oh great, DirectX 11. Despite the fact that DirectX 10 hasn't even been out that long, I will soon have to buy ANOTHER new graphics card to use it.
I'm tired of this "Graphics > Gameplay" formula that has torn through the gaming industry over the past few years.[/QUOTE]
This is why Blizzard is an awesome company.
[editline]07:37PM[/editline]
[QUOTE=FunnyGamer;20138647]AT this point the thread has become a debate on science/pseudoscience as well as a debate on the vast/minute differences between Dx10 and Dx11.
You're all dumb for busting a nut over this shit.[/QUOTE]
You're the dumb one being so incredibly ignorant judging people without knowing them.
[QUOTE=Wheeze201;20140765]This is why Blizzard is an awesome company.[/QUOTE]
Except for the whole "pay $50 per episode of Starcraft 2," sacrifice gameplay for subscribers formula with WoW, and complete abandonment of everything they once stood for because of the acquisition of them by Activision then YES.
Yes, Blizzard is the best company on the planet and I want them to rape my face.
[QUOTE=Wheeze201;20140765]You're the dumb one being so incredibly ignorant judging people without knowing them.[/QUOTE]
I don't need to know people to judge them when they're creating such petty arguments as to why creating models with a billion poly's is better than Tessellation, as well as the difference/lack of differences between Dx9, 10, and 11.
[QUOTE=FunnyGamer;20140908]I don't need to know people to judge them when they're creating such petty arguments as to why creating models with a billion poly's is better than Tessellation, as well as the difference/lack of differences between Dx9, 10, and 11.[/QUOTE]
If you don't like it, get out of the thread. I don't blame you for not being interested in this stuff, but some people actually are. We're trying to have a serious discussion here. If you think these things are unimportant, that's just your opinion, no need to shit up the thread because of it.
[QUOTE=pebkac;20142884]If you don't like it, get out of the thread. I don't blame you for not being interested in this stuff, but some people actually are. We're trying to have a serious discussion here. If you think these things are unimportant, that's just your opinion, no need to shit up the thread because of it.[/QUOTE]
I'm actually criticizing the level to which they take the discussion, which is very asinine.
Each party is defending their side so fervently it's as if their life depends on it, which kind of defeats the purpose, almost turning it into a cockwave of facts that nobody listens to.
I'm interested in this stuff too, but not on such a fanatic level (I mean, 1 billion poly models? Really? It'd be much more easier to have a single large poly model with a displacement map and let the Tessellation handle it all, as opposed to a software LOD system, so there's a lot less load on the CPU and RAM.)
Also btw, DX11 is partially compatible with DX10.1 hardware, and DX10 hardware I imagine as a consequence considering the nearly nonexistant differences between the two.
I love how people in this thread are all "[B]OMG DX11 is the devil! Kill it! It's trying to outdo OpenGL! It has new effects! It fucked my mom![/B]"
Seriously, guys. It's not like they're forcing you to buy this. It looks cool, some people, like me, like things that look cool. Man, you guys will start an internet bitch fest over nothing. :colbert:
Well this video is very old.
Also this is less resource intensive and allows for better lod's because you got tesselation up close and normal from afar.
All this is done on the card to so it can be utilized on models and other objects that have action and shit computed by the processor.
When I ran the benchmark you had to choose whether to run in DX10 or Dx11 when you started it. You can't flip ingame. All he was doing was turning on tessellation in game, but while in DX10.
[QUOTE=Muscar;20106651]What? No, real shadows are sharper when the object is closer to the shadow, and blurrier the further away it gets.[/QUOTE]
Actually, no, not always.
It's depended on the size of the light source, the distance between the subject to the light and the subject to the shadow destination. if you have a point light of a single unit. All shadows will be effectively pixel thin, since there is no change in the light path.
The example photos work that way because the light source is large and the closer the subject is to the light source the larger the angle between the left side of the light source, the subject, and the right side of the light source.
[QUOTE=PelPix123;20123551]Nope. Shadow blur is caused by the area of the light and the way the photons are emitted from that spread of light. The blur you are seeing is effectively separate shadows from an infinite amount of mathematical points emitting photons over the area of the light.
If the light was a point with no area, the shadows would not be blurred.[/QUOTE]
That.
Edit: Example:
[IMG]http://i46.tinypic.com/xeli76.jpg[/IMG]
Look at this rock I found at the base of this wall! Ten... eleven (repeat x5) Look at this shingle! Ten... eleven.
I think that they purposely made 10 look bad, much like that webpage microsoft did comparing Halo graphics to Crysis or whatever. That slanty staircase could be done much better in DX10
This thread is very informative on shadows. I like.
[QUOTE=Brt5470;20147344]When I ran the benchmark you had to choose whether to run in DX10 or Dx11 when you started it. You can't flip ingame. All he was doing was turning on tessellation in game, but while in DX10.
Actually, no, not always.
It's depended on the size of the light source, the distance between the subject to the light and the subject to the shadow destination. if you have a point light of a single unit. All shadows will be effectively pixel thin, since there is no change in the light path.
The example photos work that way because the light source is large and the closer the subject is to the light source the larger the angle between the left side of the light source, the subject, and the right side of the light source.
That.
Edit: Example:
[IMG_THUMB]http://i46.tinypic.com/xeli76.jpg[/IMG_THUMB][/QUOTE]
Crysis [I]kind of[/I] does this with the terrain.
A shadow from a mountain 500 metres away won't create a clear shadow at all, the blur wil be multiple meters wide.
While, if you just made a wall of terrain, it'd have a clear shadow along the bottom.
[QUOTE=paul simon;20152237]Crysis [I]kind of[/I] does this with the terrain.
A shadow from a mountain 500 metres away won't create a clear shadow at all, the blur wil be multiple meters wide.
While, if you just made a wall of terrain, it'd have a clear shadow along the bottom.[/QUOTE]
No, that's just the engine downsizing the shadow's texture so it doesn't rape your FPS. Same thing happens with trees.
I think that developers have really been promoting the new DX versions the wrong way. They show us pretty pictures and cool videos but for the most part it's completely unimpressive because everything we see can be made with almost any DX version. The biggest differences usually only apply to developers in that newer DX versions provide better ways of programming certain effects.
Rather than pretending like new DX versions will provide magically superior graphics, they should just say "This version of DX is better than the last. Eventually most video game developers will be using the newer version so you should consider getting a compatible card the next time you buy one."
It's fucking sickening when they do shit like compare Halo 1 to Crysis in order to trick people into buying hardware they don't really need yet.
[QUOTE=johanz;20106732]Triangle has 3 edges, quad has four edges, polygon can have unlimited amount of edges. Every game engine uses triangles as far as I know.[/QUOTE]
Triangulation.
[QUOTE=Odellus;20152278]No, that's just the engine downsizing the shadow's texture so it doesn't rape your FPS. Same thing happens with trees.[/QUOTE]
It's both. The human eye sees shadow the same as the engine portrays it to a large extent.
Zoom in with a scope and the shadows sharpen up, same goes with using binoculars in real life.
[QUOTE=Brt5470;20147344]It's depended on the size of the light source, the distance between the subject to the light and the subject to the shadow destination. if you have a point light of a single unit. All shadows will be effectively pixel thin, since there is no change in the light path.
The example photos work that way because the light source is large and the closer the subject is to the light source the larger the angle between the left side of the light source, the subject, and the right side of the light source.
Edit: Example:
[IMG]http://i46.tinypic.com/xeli76.jpg[/IMG][/QUOTE]
Can't even get your own logic right, eh? There's no hourglass shape numbnuts, or are you suggesting the shadow is millions of kilometers wide far out in space?!
Once more, don't disagree if you don't disagree. Any shadow that is even slightly blurred will be moreso further away. And since point sources do not exist in the real world that is all shadows.
Don't play dumb to troll. Or be dumb for that matter.
[editline]06:32PM[/editline]
[QUOTE=Odellus;20152278]No, that's just the engine downsizing the shadow's texture so it doesn't rape your FPS. Same thing happens with trees.[/QUOTE]
No, even upclose the shadow from terrain is deliberately blurred according to distance.
Sorry, you need to Log In to post a reply to this thread.