Displacement Mapping Question

[pav]

ok, i'll simplify my question.

is Displacement mapping possible on a X800XT? if so, can anyone please give me any resources at all to acheive that? all i find in google is reference to either modeling software displacement mapping, or nvidia's "Look at our supreme VS3.0 ~_~" displacement mapping.
is there really no way to implement it on a X8** hardware?

 

[ShaidarHaran]

Apparently one of the features SM3.0 adds is displacement mapping, so I doubt you'd be able to get that function to run in hardware on an X800-class card.

 

[Cassini]

ok, i'll simplify my question.

is Displacement mapping possible on a X800XT? if so, can anyone please give me any resources at all to acheive that? all i find in google is reference to either modeling software displacement mapping, or nvidia's "Look at our supreme VS3.0 ~_~" displacement mapping.
is there really no way to implement it on a X8** hardware?

Doesn't look like it is possible, although "virtual displacement mapping" or "parallal mapping" which are the same are possible on SM2.0 hardware.

Real displacement mapping is a new feature of SM3.0.

 

[razor1]

It is possible, first pass you can do a texture look up though a pixel shader, second pass displace, it is a bit slow around a 50% overhead. Also if tesselation is involved another pass is required so its more like 75% overhead or more since now you have an increase of polygons.

http://www.cgl.uwaterloo.ca/~krmoule/dmap_gi2002.pdf
This was done on GF 3 hardware part of the algo was in software pure hardware accelerated adaptive tesselation check nvsdk 9.1 there is an example there in Cg.

 

[Humus]

Apparently one of the features SM3.0 adds is displacement mapping, so I doubt you'd be able to get that function to run in hardware on an X800-class card.

What SM3.0 adds is vertex texture access. This is not the same as displacement mapping. To get true displacement mapping you'd need tesselation support as well.

 

[pav]

i read a presentation by Natalya Tatarchuk on Parallax Occlusion Mapping the results of which seemed "close enough".

does anyone know of more resources on the subject?

 

[ShaidarHaran]

What SM3.0 adds is vertex texture access. This is not the same as displacement mapping. To get true displacement mapping you'd need tesselation support as well.

I'll take your word for it rather than that of some PR/marketing troll
Thanks Humus.

 

[razor1]

i read a presentation by Natalya Tatarchuk on Parallax Occlusion Mapping the results of which seemed "close enough".

does anyone know of more resources on the subject?

Check nvsdk the lastest one there is a shader that with parrallex self shadowing. it works well but the self shadowing is a bit rough. The shader can be modified to give a more softer shadow.