>and on another note, i know we've been over this a billion times in
>lecture, but ... can you please post the solutions to prelim2
>(particularly the last question, concernig ray-plane intersection) or
>make hardcopies available somewhere (maybe they are in 303 already?)?
>thanks,

the prelim2 solutions are outside upson 303 already.

(now, back to your original questions...)

>I'm a bit confused about how we should do zbuffering for spheres now. I
>understand that for the polygon meshes it is easy to project vertices
>and interpolate for zbuffering, but since the spheres are defined by
>implicit surface formulas, are we now actually shooting rays at a
>sphere? 

yes.

>	as you said in a previous post, do you mean that when we are
>checking to find the new boundary on a neighboring row, that we are
>shooting rays at the sphere through each screen coordinate (ie 1 per
>pixel if no supersampling) until we miss, then moving on to the next row
>and repeating over the height of the sphere? 

that's basically right, except that "height" is not such a simple
concept: keep processing rows until you no longer hit the sphere.
also, you don't want to touch all pixels on each row, but some subset
of pixels likely to touch the sphere -- a good heuristic is to look
near the pixels touched in an adjacent row.

>					      or am i just being dumb and
>missing the way to "project" the equation for the sphere onto the view
>plane and use that to check to see if a screen coordinate is inside the
>projection of the sphere?

there might be a simple way to project, but nothing immediately came
to my mind when i briefly thought about it...

-- 
Thomas Yan