Hi, I want to use floating-point "rendertarget"-like things in OpenGL. I found some samples for DirectX, but nothing for OpenGL seems to be able to do this. A float Pixelbuffer seems a solution, but this won't work nice either. Even on a NV GeForce 6600, I only get 2-3 fps with resolutions higher then 256x256. Has anyone a tutorial/manual/link that explains how to use floating point targets in OpenGL? It has to work some way, because nvidia itself has already released GL-based high-precision demos - but i have not found sourcecode for them yet. (e.g. the ship-demo builds on glut). Any help is highly appreciated because i spent a lot of time in searching a solution... :D Greetings, T.

shouldnt be as hard as you say, especially not on NVIDIA hardware.
have you reviewed these extensions:
- GL_ARB_texture_float
- GL_NV_float_buffer

I'm also surprised there's no example code for this on developer.nvidia.com, I swear I've seen some..somewhere....

Cheers
-Danu

Well.. I should add that I have to use a Linux-environment. I found some SDK-Demos (especially hdr and 16bit floating point blending and filtering) that use floating point pbuffers - but run only on windows systems. Could this be a Linux-Limitation? Well, I can initialize float pbuffers - but they are awsome slow, seems like software-calculated. I also have the float-Extensions and so on...

But surely, it works on linux boxes, too... perhaps theres something that I missed.... It would be of great help if the primary display-context could be floating point, but this has du be clamped rgb as it seems (e.g. there is no GLUT_FLOAT and so on).

Thanks,

T.