Quote:

Original post by skow Well I figure I would have to render that step 1 to the back buffer as I am then going to blend over it with step4. Also, why not use the ARB_TEXTURE_RECTANGLE? Thanks

omfg, theres and ARB texrect extension :| ???

I get the loading screen, then it starts to load, then a window pops up and says "Couldn't fetch valid FP profile," then another window says "Ahh, crap, screwup...".

I'm running a 64 MB Radeon 9000 on a Dell Inspiron 600m, with 512 MB RAM. I'm using one of the more up-to-date Omega drivers.

-Gauvir_Mucca

Quote:

Original post by silvermace omfg, theres and ARB texrect extension :| ???

Indeed there is, although I'm yet to use it. I assume it works just like the NV one.

So I'm assuming I'll just have to choose from :
- using glCopyTexSub and draw the scene once and use glCopyTexSub and render to a reduced size pbuffer

or

- drawing it twice once at full size to the backbuffer and a second time to the pbuffer with viewport 1/2 the size?

Ideally I would like to just draw it once and have glCopyTexSub create a texture 1/2(height)x1/2(width) the size of the buffer. Is there any way this is possible?

why not use ARB_texture_non_power_of_two if you can (you need a fallback though), it uses a bit less memory than ARB_TEXTURE_RECTANGLE and it hs no real limitations.

Intel P4 2.8GHZ HT
Radeon 9800 Pro

[Fullscreen]
No fx: 340 fps.
fx: 70 fps.

Quote:

Original post by skow Quote: Original post by silvermace omfg, theres and ARB texrect extension :| ??? Indeed there is, although I'm yet to use it. I assume it works just like the NV one.

Right. Taken from the spec :

Quote:

Backwards Compatibility This extension is semantically equivalent to EXT_texture_rectangle and NV_texture_rectangle. The tokens, and name strings now refer to ARB instead of EXT or NV. Enumerant values are unchanged.

Quote:

Original post by lc_overlord why not use ARB_texture_non_power_of_two if you can (you need a fallback though), it uses a bit less memory than ARB_TEXTURE_RECTANGLE and it hs no real limitations.

ARB_npot is supported by a very short range of graphics cards, namely the new generation (GeForce 6 and-the-like).