fbo problems

you're my hero. never thought it could be related to that. anyways, i finally got it working now. thanks a lot.

yet another problem i ran into. so i got rtt set up properly and implemented a type of motion blur with it, looking like this:

http://www.fichtenmoped.at/maya_knight/cube_blur1.jpg

so far all i did is decrease opacity. now i'd like to start messing with a gaussian blur filter on the screen rendered to a texture. for now it's okay for me to just go with the slow texture copy functions to get the texture into the main memory and then do the gaussian filter there. but after plugging in this piece of code:

GLubyte tex_data_[TEX_SIZE][TEX_SIZE][4];
glBindTexture(GL_TEXTURE_2D, tex_]); //tex_ is the whole screen, rendered to this texture
glGetTexImage(GL_TEXTURE_2D, 0, GL_RGBA, GL_UNSIGNED_BYTE, tex_data_);

//gaussian filter to be inserted here later.
//empty for now.

glTexImage2D(GL_TEXTURE_2D, 0, GL_RGBA, TEX_SIZE, TEX_SIZE, 0, GL_RGBA, GL_UNSIGNED_BYTE, tex_data_);

the result looks as follows:

http://www.fichtenmoped.at/maya_knight/cube_blur2.jpg

I didn't perform any manipulation of the texture at all yet. All i did was get the texture (which holds the current frame) into the main memory and then put the very same data back. This texture is then just drawn to the screen with a plane and an orthogonal view. So in theory it should look like the first pic, but something seems to be going wrong.
Could it be that glGetTexImage() is just too slow at copying, producing these results? It's still weird though that there's this pattern in the blur.

Concerning your new problem, I think it is not only a problem of speed since the colors of the cube change over time (or did i misunderstood something?).
Concerning your approach for post processing the texture, i'm not sure if it is the best way to do this, but i think you could do it on the gpu. Bind your texture, set viewport size to match your texture size, render a quad on screen and in the pixel shader, apply the filter, using gl_FragCoord.xy as texture coordinates (if using GLSL) to read your texture.
If you don't want to use shader, my post is definitely useless, sorry.

It looks to me like the red and blue channels are getting swapped during the copy.

When you copy out the first time red becomes blue and yellow becomes cyan. Then on the next frame, you copy out again and red/blue and yellow/cyan become inverted again and restore back to their original colors.

Not sure _why_ that would be happening, but that's what it looks like. :)

Edit: Now that I think about it, you might want to try glReadPixels instead of glGetTexture... I've never actually used glGetTexture though.