Radeon 9700 & PBuffers

I’m using Pbuffers for a render-to-texture effect in a game we’re working on, but I’m having problems with it on the 9700 card.

In our game, we render the scene to a texture (the pbuffer) and then use that texture for some blurring effects by layering it multiple times on the screen. It works perfectly on a GeForce4, but whenever I try testing it on the 9700, the render-texture acts like it’s getting lost when it’s selected a second time. I can use the render-texture to render as many polys as I want as long as I don’t call glBindTexture on it more than once. Once I call it a second time, the polys show up white (almost as if I had passed in a null pointer) until I regenerate the render-texture. Just to make sure it wasn’t getting lost somewhere else in the code, I tried calling glBindTexture twice in a row and rendering a poly using the texture and it never shows up. If I remove the second glBindTexture, it does show up.

Has anyone else experienced this? I tried it on two different computers with a 9700 installed and had the same problem on both. But on the GeForce it works fine. I there an attribute that I may be leaving off when creating the pbuffers that could cause this, or is there something else I’m not thinking of?

Thanks for any help,
Marc

There are some bugs with PBuffers and WGL_ARB_r_t on the 8500 and 9700. Most of them have been worked out in later drivers though.

Which driver set are you using?

Try turning off FSAA and see if it helps. I need to turn it off in the current drivers for rendering to texture to works as supposed.

I just ran into this problem 5 minutes ago. I have two color buffers of a pbuffer bound to textures. But before I need them, I bind another texture ( standard ) and perform a glCopyTexSubImage2D operation. This is where it breaks down. If I don’t use glCopyTexSubImage2D but rather a second pbuffer for this step, everything works. Note: the copy step is to a standard 2D image, not a bound pbuffer color buffer. On GeForce3 both methods work as expected.

[This message has been edited by PH (edited 12-14-2002).]