a little "strategic" help

hi dudes,
i am needing a bit of strategic help from some advanced opengl programmers:
I wand to build a little engine, able to work with env maps,multitextures, realistig lighting/shadows all in realtime.
Actually my boss suggested to use a radiosity model + an hdri enviroment, but after some preliminary studies i just came up thinking that a radiosity engine just won’t make any good performance realtime. (+ it wouldn’t work properly with reflective objs).
Now i was thinking about vertex shadig/pixel shading, maybe using OGL 2.0 with the shader language features. The pros are tha if i’ve read all well it should do all i need (and much more) and that seems to be the “way of the future” in 3d graphics. The little flaw is that it looks to me quite hard to understand, manage, use in all day work.
Now i am asking to you, oh munificent OpenGL gurus, what kind of choice should i make? Does opengl 2.0 & shader lang are really SO cool? Reading the 3dLabs stuff the seem awesome, but after all that is their product
thank you, regards
The Gunslinger

isn’t a lot of stuff the same, like loading mesh data, object data, textures, models, etc., so that you could just start with your “litlle” (doesn’t sound like it will stay little) engine and improve it step by step, first use very stupid vertex lighting and then improving?


(disclaimer) I won’t bother to point out all sorts of stuff. Oh, and I’ve got a bad flu.

OpenGL 2.0 isn’t out yet.
Vertex programs, at least, are quite widely avaliable and really easy to use.

I think one important decision you need to make is what hardware you’re targeting. Take it from there.

Radiosity is not for real time. It can be precomputed for a static solution in real time though. There are also things like radiosity maps but they are a different thing really and very limited.
Radiosity is for diffuse lighting. That doesn’t mean it doesn’t work with reflections, just that you need a different solution for them.

Reflections, you’re bound to do with some kind of env-mapping. The common methods are cube-mapping, dual-paraboloid mapping and sphere-mapping (this latter generally being view-dependant).

There are tons of lighting solutions and hybrid solution are well in order.

The general trend these days is to do per-pixel lighting with bumpmappiing support and shadows maps or stencil shadows. There’s quite a few approaches to the former and you will commonly want more paths for different hardware.

It’s such a general question that I could go on for a lot longer and come up with a long list of things to google for but I’ll leave it here for now.

thanks for the little hints:

madoc sorry 4 your flu (che culo, sotto le feste ).

i am sniffing a bit into the wonderfull world of extensions, ARB_HERE, ARB_THERE…
it looks kinda nice stuff, even if so far i am not really understandig a lot.
Guess the solution is all the same, just getting myself to work and test some x pixel lighting stuff.