LOL..this cracked me up: “But aren’t G-buffers getting bigger all the time?” Yes, but that’s NOT necessarily a good thing. 😉

Seriously, yes, in the long term we can expect the cost of the G-buffer to go down as we get faster busses, more ROPs, more VRAM, bla bla bla. But increasing the size of the G-Buffer is a fairly drastic and non-incremental thing – it’s a system-wide tax, so I’m not looking to push it and push it and push it. The _coding_ of specular tinting is pretty trivial – it’s the G-buffer cost that’s an issue. (And once we have it, we’ll end up with specular-tinted content.)

I am also _not_ convinced that specular tinting is the best use of G-Buffer space within the bigger scope of ‘materials’, something that unfortunately G-Buffers don’t handle with a lot of grace. I want to understand what else we want to spend g-buffer budget on before spending that budget.

PS. I’m not a nerd, I’m a geek. Geeks get it done! 😉

To be sure. But aren’t G-buffers getting bigger all the time? Where will we be in two years? Not quibbling. I guess it depends on how easy RGB tinted specularity is to code, Ben. Processing it might be a bridge too far now, but when we have the advantage of 20/20 hindsight, it might be seen in a significantly different light. No pun intended. Thanks for not closing the door.

Reflections: some day…yes.

http://en.wikipedia.org/wiki/Specular_highlight

Not sure. We might get in one, we might get in two. If the first one is dead, we’ll definitely get two. 😉

Not much hope. We’ll see, but not much. If ANY part of the sim uses tinted specular hilites, we have to pay for it in the g-buffer.

That is greate!!!