Revealing The Power of DirectX 11

That looks really sweet!
Thanks!

Will everyone end up using "the same" tessellation code? If the modelling tools are pretty much common across the industry is there going to be much variation in tessellation algorithms in games?
My guess is that they will be very similar. This sort of rendering is pretty stable as far as algorithms are concerned - I'm sure new ones pop up now and again, but for the most part artists know how to use certain tools and they're happy with them. Provided those can be efficiently implemented in D3D11 then why rock the boat?

The differentiator I can imagine is any additional control features - you definitely want/need control over creases and sharp edges if you're using HOS and these may well need to be implemented in a custom manner depending on the art tools in use...

Of course, there will be developers who want to do something different :cool:

I suppose there are interactiions between tessellation and types of rendering pass, e.g. shadow map generation and the use of silhouette-edge-refinement tessellation.
Yup, in my opinion it was the adoption of stencil shadowing (=heavy on geometry) that killed D3D8 HOS (aka ATI's TruForm) but with shadow mapping being more common now its less of an issue. Those silhouette algorithms look very nice - thats one case where a simple algorithm can really improve the visual quality and eliminate one of the oldest and most common "uglies" in raster graphics!

I really want a d3d11 gpu
Join the queue :smile: I'm doing all this work on the reference rasterizer which isn't exactly interactive. Makes me glad I have a quad core @ 3.1ghz...


Cheers,
Jack
 
Do you guys know how much of DX11 can be brought to 360? I know there should be a lot of limitations but do you guys think whether it is possible to bring the new shaders in some form? (i.e. compute shader)

On a side note, is it possible to emulate the fixed tesselators in ATI's DX10 class hardware and in 360 with the new tesellation system?
 
Do you guys know how much of DX11 can be brought to 360? I know there should be a lot of limitations but do you guys think whether it is possible to bring the new shaders in some form? (i.e. compute shader)

On a side note, is it possible to emulate the fixed tesselators in ATI's DX10 class hardware and in 360 with the new tesellation system?

If I've understood it correctly, the DX11 tesselator is superset of the 360/R600 and RV670/770 tesselators (or what was that point where they tweaked it a bit? 670 or 770?`)
 
I've been doing a lot of work since the MVP Summit in March on D3D11 Tessellation, all of which is posted in my developer journal. In particular I posted a big diagram and explanation of the new pipeline stages - colour coded with arrows and everything :smile: - you can view that here.

Nice!! This sure beats all those vague powerpoint slides :) But I'll admit I'm still not fully clear on the relationship between the weights output by the tessellator and the HS control mesh output and how they're used together to generate new vertices in the DS.
 
The tesselation seems pretty haphazard to me, none of the "classic" metrics are being optimized ... neither screen space extent nor curvature. How good exactly is the adaptivity we can expect in actual games, with no offense intended I do hope better than this :/
 
Do you guys know how much of DX11 can be brought to 360? I know there should be a lot of limitations but do you guys think whether it is possible to bring the new shaders in some form? (i.e. compute shader)

On a side note, is it possible to emulate the fixed tesselators in ATI's DX10 class hardware and in 360 with the new tesellation system?
Xenos doesn't have a compute shader concept so its traditional GPGPU for processing.

I think possible to get close to Xenos' tessellation algorithm with DX11, but there might be some differences. Basically I'm not sure on that.

DX11 Tessellation and 360 tessellation are most different in 360 doesn't have a hull shader. 360 implements adaptive tessellation with a 2 pass approach. The first is like a geometry shader that doesn't amplify. The second pass receives the tessellation factors as inputs and the tessellated points are run through the VS which works like the DS in DX11.

So in essence it's not a straight forward port between DX11 <-> 360.
 
Back
Top