randycat99
Veteran
Would that make it more of a vertex shading operation than a pixel shading operation then?
Bungie Halo2 illustration of dot3 bump map modeling
- Thread starter duffer
- Start date
Displacement mapping will only be seen once there are a reasonable amount of mid to low end videocards with displacement mapping, that is, DX9.
The most promising thing in this regard is that ATi is already planning very inexpensive DX9 parts (Radeon9500) and nVidia seems like they'll have NV31 (MX version) very soon after NV30. Which means that we won't have to wait quite as long as from the DX7 of GeForce to the DOOM3 of next year.
Even then, we are unlikely to see anyone use displacement mapping until XBox2, or the next Carmack engine after DOOM3, whichever comes first.
The most promising thing in this regard is that ATi is already planning very inexpensive DX9 parts (Radeon9500) and nVidia seems like they'll have NV31 (MX version) very soon after NV30. Which means that we won't have to wait quite as long as from the DX7 of GeForce to the DOOM3 of next year.
Even then, we are unlikely to see anyone use displacement mapping until XBox2, or the next Carmack engine after DOOM3, whichever comes first.
I think we'll see a few games before that, just as there was DOT3 bump mapping (Giants) and stencil volume shadows (Blade of darkness) before Doom3.
The interesting question about displacement mapping is shading.
Should it be performed in the vertex shader with custom vertex lighting routines? But how would one use gloss maps and such there? Or should it be per-pixel shading using vertex normals instead of normal maps? Or the combination of the two?
Also, higher polygon counts will require advanced skinning methods. No matter how cool matrix palette skinning is considered to be, it does not preserve volume and cannot produce bulging and such. Natural-looking creatures would require more complex tools like muscle simulation or at least a lot of additional deformers...
The interesting question about displacement mapping is shading.
Should it be performed in the vertex shader with custom vertex lighting routines? But how would one use gloss maps and such there? Or should it be per-pixel shading using vertex normals instead of normal maps? Or the combination of the two?
Also, higher polygon counts will require advanced skinning methods. No matter how cool matrix palette skinning is considered to be, it does not preserve volume and cannot produce bulging and such. Natural-looking creatures would require more complex tools like muscle simulation or at least a lot of additional deformers...
Only problem is performance... Geometry amplification ?
Well displacement map do increased the geometry of your model. Is that what he meant ? Maybe displacement map can increased your tesselation in unpredictable way, that can be the case if it was quad, generating those non-planar stuff, but I think you can get more control in tri. But still displacement map is also a fill gap solution like normal map, until you can just shove the high poly model down the bus.
Also it wouldn't work in his Doom 3 engine, since the shadow is calculated in CPU.
Well displacement map do increased the geometry of your model. Is that what he meant ? Maybe displacement map can increased your tesselation in unpredictable way, that can be the case if it was quad, generating those non-planar stuff, but I think you can get more control in tri. But still displacement map is also a fill gap solution like normal map, until you can just shove the high poly model down the bus.
Also it wouldn't work in his Doom 3 engine, since the shadow is calculated in CPU.