Re: [Algorithms] Re: Portal VS BSP

[Alen L. <ale...@cr...>]

> > (hierarchical) view frustum culling algorithm (for fixed,
> > specialized scenes)" in the original dPVS manual (at least we
> > did in the subsequent journal paper), the same way I regard

Is this journal paper public?

> > back-face culling to be an "occlusion culling algorithm for
> > closed, watertight objects with limited camera motion" (think
> > a moment about the latter :)).
>
> Strangely, I don't tend to think of backface culling as a visibility
> algorithm, but more as a rasterisation rule that eliminates invalid drawing
> of the object's surface which is, after all, a polygonal representation of
> the surface of a volume. The back faces of these polygons are not part of
> that representation (it is only supposed to work looking from the outside)
> and so should be removed.

For a rough classification at the top level, I'd give this:

VSD can be classified:
A) by geometric exactness into
  1) culling,
  2) clipping,
B) by type of (in)visiblity into:
  1) backface culling/clipping - what faces away from viewer,
  2) frustum culling/clipping - what is outside of viewer's FOV,
  3) occlusion culling/clipping - what is hidden from the viewer by some
nearer object.

In this notation, "backface culling" would be a subclass of "visibile surface
determination" algorithms, but it is distinct from the "occlusion culling"
class.

Cheers,
Alen