Correct me if my impressions are wrong here. Isn't this whole situation analogous to the DX 8 sm 1.1 vs 1.4? Nvidia argued that there was nothing that could not be done in 1.4 that could not be done in 1.1. So basically we have a reversal of roles here with ATi and nVidia. If so sm 3 should come off in a similar manner as 1.4 did or no? I suppose it will all come down to developer support if that is the case. Also if nVidia is not able to hold the high end in terms of performance I don't think that boasting sm 3 will help them in terms of maintaining market share. Brandishing ps 1.4 certainly didn't help ATi much with the sales of the R200.
[H]'s take on SM3 pros/cons
- Thread starter Sanctusx2
- Start date
As far as I'm concerned I think that THE biggest advantage of PS3.0 over PS2.0 is that this model is not "fragmented" with things like PS2.0, PS2.0a PS2.0b. At least with PS3.0 you now for sure that you have :
-512 instructions
-no restriction on these 512 instructions (ALU instructions or texture instructions posibly dependant)
-dsx dsy instructions
-centroid sampling
-MRT
All of these were available under different profiles for PS2.0 but it was so messy that developers chose to stick to straight 2.0. PS3.0 is giving a clear target for developers on what features to expect.
-512 instructions
-no restriction on these 512 instructions (ALU instructions or texture instructions posibly dependant)
-dsx dsy instructions
-centroid sampling
-MRT
All of these were available under different profiles for PS2.0 but it was so messy that developers chose to stick to straight 2.0. PS3.0 is giving a clear target for developers on what features to expect.
radar1200gs
Regular
What follows is speculation on my part.
We don't yet have the FarCry patch that enables SM3.0 support so we don't know exactly what it will do yet.
My guess is that SM2.0 hardware will make use of offset mapping using the pixel shaders to emulate displacement mapping, while SM3.0 will actually do true displacement mapping in the vertex shader. Of course, we won't know until the patch is actually released.
We don't yet have the FarCry patch that enables SM3.0 support so we don't know exactly what it will do yet.
My guess is that SM2.0 hardware will make use of offset mapping using the pixel shaders to emulate displacement mapping, while SM3.0 will actually do true displacement mapping in the vertex shader. Of course, we won't know until the patch is actually released.
radar1200gs
Regular
Care to elaborate why you feel they won't?AndrewM said:
DW Fan!!!!!
Newcomer
Well if you read the beginning of this thread, DemoCoder doesant think so either and he loves NV. 
radar1200gs said:
It requires quite a bit more work than what people think, on the content creation side. The pixel shader offset/parallax/virtual displacement bump mapping technique is much cheaper (for the same result) and gives about the same visual quality.
I think the [H] article was pretty crappy. The bit where Brent talks about lighting not affecting the surface properly because it's "flat", is totally incorrect (unless you're doing some things in the vertex shader).
Doomtrooper
Veteran
What is the development graphic card that is being used for TGC engine Andrew McGregor ? What HLSL is being used, I know you have a love for CG as I see your posts on Sumea Are you using both ?
Are you using both ?Doomtrooper said:What is the development graphic card that is being used for TGC engine Andrew McGregor ? What HLSL is being used, I know you have a love for CG as I see your posts on Sumea
Geforce3 is our min hardware spec. I use GLSL and Cg (for various reasons)
overclocked
Veteran
I´ve not read what you all been said but just looking at the launchvideo again/and been away from 3dGfx for app 10weeks The guy from Crytec must made nVidia mad as everytime the nV guy said "this is done with the SM3 in just 3 weeks in Farcry" he EVERYTIME said SHADERMODEL2. AND SM3.
The guy from Crytec must made nVidia mad as everytime the nV guy said "this is done with the SM3 in just 3 weeks in Farcry" he EVERYTIME said SHADERMODEL2. AND SM3.
And just to keep people off-balance, a serious and on-topic question from me!
From Brent's article I gathered that displacement bump-mapping actually does improve visuals when the textures are viewed up close....is displacement bump-mapping one of the bits of SM 3.0 that the R420 is rumored to support? It sounds like the important eye-candy bit, so I thought I'd ask.
I keep hearing it bandied about that the only thing in SM 3.0 that ATi won't support is branching or full-precision, would either of these be required for displacement bump-mapping? :|
From Brent's article I gathered that displacement bump-mapping actually does improve visuals when the textures are viewed up close....is displacement bump-mapping one of the bits of SM 3.0 that the R420 is rumored to support? It sounds like the important eye-candy bit, so I thought I'd ask.
I keep hearing it bandied about that the only thing in SM 3.0 that ATi won't support is branching or full-precision, would either of these be required for displacement bump-mapping? :|
digitalwanderer said:And just to keep people off-balance, a serious and on-topic question from me!
From Brent's article I gathered that displacement bump-mapping actually does improve visuals when the textures are viewed up close....is displacement bump-mapping one of the bits of SM 3.0 that the R420 is rumored to support? It sounds like the important eye-candy bit, so I thought I'd ask.
I keep hearing it bandied about that the only thing in SM 3.0 that ATi won't support is branching or full-precision, would either of these be required for displacement bump-mapping? :|
No neither are required to do displacement mapping (you can leave the bump part
). But from the ps2b target profile I can see another limitation of ATI next chip : even if the number of instructions is up from 96 to 512 there's still a limit on the number of dependant texture reads you can do. I guess it comes from ATI's implementation (F-Buffer based ?).I usually like Tim Sweeney, but he has managed to confuse the 3d enthusiast community quite well when he decided to call offset bump mapping as virtual displacement. Most people only remember the "displacement" part and already I've seen people telling things like UE3 is pushing 300 million polygons...
Edit: forgot the important part
So there's displacement mapping, which means that you actually move the vertices of the 3D object around in 3D space. It obviously needs a lot of geometry to look right, at least 1 vertex / texel (so an 1K texture needs a million vertices already!), but the best looks are achieved if you tesselate the model until each triangle is at least as small as a pixel.
Then there's offset bump mapping, which is an advanced form of bump mapping, and which Tim decided to call virtual displacement mapping. It basically needs a height map and a normal map, and a few extra isntructions, so it's not much beyond standard bump mapping.
Edit: forgot the important part
So there's displacement mapping, which means that you actually move the vertices of the 3D object around in 3D space. It obviously needs a lot of geometry to look right, at least 1 vertex / texel (so an 1K texture needs a million vertices already!), but the best looks are achieved if you tesselate the model until each triangle is at least as small as a pixel.
Then there's offset bump mapping, which is an advanced form of bump mapping, and which Tim decided to call virtual displacement mapping. It basically needs a height map and a normal map, and a few extra isntructions, so it's not much beyond standard bump mapping.
