"SharkD" <nomail@nomail> wrote:
> Heh. I tried reading the tutorials you linked to and have to admit I have no
> idea what is going on. Are you rendering an initial image and then wrapping it
> around the scene in a sphere? Wouldn't this in effect limit the possible shapes
> of the scene to something that can fit well inside a sphere? The scene I have
> rendered is rectangular, and there are few or no objects of interest outside
> this rectangle.

The image you posted looks like a complete 360x360 degree spherical image. If
so, this could be wrapped on a large world-sphere and be used as an
environment. The spherical camera projection works the same way as the image
mapping, so it doesn't matter what your environment scene contains, it will map
to the sphere without distortion. (The top edge of the image is the zenith (+y),
the middle row is the equator and the bottom edge the nadir (-y). The left edge
meets the right edge at  a meridian that passes through the +x axis, by
default).

Using this environment map would allow you to render, say, a close-up of a
reflective object on a table, with reflections that would be indistinguishable
from actually having the environment scene around it. If the environment were
an hdr image, you could use radiosity and allow the environment-mapped sphere
to illuminate your subject, again, in a way that would be identical to having
the actual scene there.

The advantages of this technique are faster rendering (intersection tests and
lighting calculations on the environment objects are cut to simple intersection
tests with a single sphere), and you can share the environment map without
having to supply the complete environment scene.

Bill

Post a reply to this message