Hello all,
A couple friends and I are designing a rendering engine and we ran into =
a problem. The issue is this. When you are rendering meshes with =
different effects on them, how do you generalize the rendering? When we =
looked at it, it seemed like the best solution was a base class =
DrawableMesh, with various subclasses that do different things, such as =
a BumpMapMesh, a LightMappedMesh, etc. Each one of these subclasses =
would know how to set up the rendering for themselves based upon how =
many textures were allowed per pass on that card, and the various caps =
of the card. These meshes would have their verteces, textures, and =
shaders abstracted through a manager that handles reference counting for =
those objects, and that sort of thing.
The main issue we were having is the huge amount of subclasses required. =
I mean, to us it seemed you would wind up with a class for every =
permutation(combination?) of effect. e.g. you could conceivably(don't =
know *why*, per se) wind up with a =
LightMappedBumpMappedSpecularMappedSkinnedMesh, and that would have to =
reimplement the aspect of light mapping, bump mapping, specular mappng, =
and skinning. This seemed...wrong..to us. Does anyone have any better =
ideas?
This does seem a bit general for this list, so I'm ccing the GD-design =
and GD-general lists..please reply there if you feel it more =
appropriate.
Apologies for a duplicate if any...i sent it from the wrong accoutn at =
first and the list servers didn't like that.
Any help greatly appreciated.
Dan
|
