water bump maps
Author
I know the that most Nvidia cards can''t seem to handle the water bump map feature but will the Geforce 3 handle it? I just got one and I get an incredibly weird distortion for the water. What settings, if any, will this work right on?
Thanks,
Rendar
I recently asked Nvidia about this. Here is the response I got and my original message:
---------------------------
The issue is a HW limitation, but it''s not as bad as it might seem at first.
What the GF3 can''t do is the following :
tex t0 // fetch dU dV
texbem t1 // Look up ( u + dU ) / q ( v + dV ) / q
it can do this :
tex t0 // fetch dU dV
texbem t1 // Look up ( u / q ) + dU ( v / q ) + dV
Since dU and dV tend to be small, this is a pretty good approximation. The
one case where may fail is looking through the wrong side of a single sided
piece of bumpy glass, the dU and dV might go the wrong way. Of course, you
could correct this via the 2x2 matrix, or by making the glass two sided.
I made a fixed version of BumpWaves that looks almost indistinguishable from
the original, just by doing the texture projection for t1''s original texture
coordinates on the CPU or in a vertex shader.
-----Original Message-----
From: Chris Egerter [mailto:]
Sent: Saturday, June 23, 2001 12:47 PM
To: Developer Relations
Subject: GF3, EMBM, and Projected textures
I read somewhere that EMBM and projected textures will not work on
Geforce3.
Some demos that use this (Bump Waves from DX8 and ATI''s Frosted Glass
http://www.ati.com/na/pages/resource_centre/dev_rel/sdk/RadeonSDK/Html/S
amples/Direct3D/RadeonEMBMFrostedGlass.html)
do not generate the correct texture coordinates.
Can you explain why these demos don''t work, and clarify if it will ever
work on a Geforce3? Can it be correctly in future drivers? Are there
any alternative methods?
---------------------------------------------------------
So right now there are 3 techniques needed to achieve the same effect depending on the card. Matrox G400, ATI Radeon and Nvidia GF3 all need different code. Right now only the Radeon works.
---------------------------
The issue is a HW limitation, but it''s not as bad as it might seem at first.
What the GF3 can''t do is the following :
tex t0 // fetch dU dV
texbem t1 // Look up ( u + dU ) / q ( v + dV ) / q
it can do this :
tex t0 // fetch dU dV
texbem t1 // Look up ( u / q ) + dU ( v / q ) + dV
Since dU and dV tend to be small, this is a pretty good approximation. The
one case where may fail is looking through the wrong side of a single sided
piece of bumpy glass, the dU and dV might go the wrong way. Of course, you
could correct this via the 2x2 matrix, or by making the glass two sided.
I made a fixed version of BumpWaves that looks almost indistinguishable from
the original, just by doing the texture projection for t1''s original texture
coordinates on the CPU or in a vertex shader.
-----Original Message-----
From: Chris Egerter [mailto:]
Sent: Saturday, June 23, 2001 12:47 PM
To: Developer Relations
Subject: GF3, EMBM, and Projected textures
I read somewhere that EMBM and projected textures will not work on
Geforce3.
Some demos that use this (Bump Waves from DX8 and ATI''s Frosted Glass
http://www.ati.com/na/pages/resource_centre/dev_rel/sdk/RadeonSDK/Html/S
amples/Direct3D/RadeonEMBMFrostedGlass.html)
do not generate the correct texture coordinates.
Can you explain why these demos don''t work, and clarify if it will ever
work on a Geforce3? Can it be correctly in future drivers? Are there
any alternative methods?
---------------------------------------------------------
So right now there are 3 techniques needed to achieve the same effect depending on the card. Matrox G400, ATI Radeon and Nvidia GF3 all need different code. Right now only the Radeon works.
This topic is closed to new replies.
Advertisement
Popular Topics
Advertisement
