--------------------------------------------
student    : asundqui
graded by  : Damian Isla
time       : Wed Nov 17  1:32:59 EST 1999
problem set: ivray
--------------------------------------------
-1 PT: NO EXECUTABLE TURNED IN
Eye ray generation            (0-2): 2
Cone, cyl, poly isection      (0-2): 1.5
Shadow                        (0-1): 1
Reflection                    (0-2): 2
Transmission                  (0-2): 2
Shading			      (0-2): 2
Supersampling and jittering   (0-2): 1.5
Statistics		      (0-1): 1
Exactitude                    (0-1): 1
total                        (0-15): 13
--------------------------------------------
notes:

Great job on a very tought problem set. I had just a few problems with
your code:

There seemed to be something wrong with polygon intersections: all
polygons are being rendered as flat black.

Your jitter was not quite right. When you jitter with super-sampling
you want to start with the regularly sampled subgrid of the type that
you create when doing normal supersampling, but then perturb each of
these rays slightly within its own subgrid.

As for the comments in your READTHIS:

2) In general, we do color blending this way so that we can cleanly
   separate specular highlights from reflectivity. In your scheme, any
   material with a specular highlight must also be reflective --
   obviously not the case for materials like plastics, brushed metal
   etc. You're right in one respect: perhaps it is fairly arbitrary to
   use the SAME value for calculating highlights and for scaling
   reflected radiance. What we're doing in the solutions is a bit
   tricky -- we're using obj->ks to test whether or not the material
   is reflective at all (and we update the weights by this value) and
   then using obj->SpecularColor to actually do the scaling of the
   reflected radiance. You're right, there's probably a better, more
   physical way to do all this.

3) You're also right that transparency and reflectivity should not
   interact in any way -- this is physically accurate as
   well. However, realize that multiplying the other color components
   by (1-transparency) would be wrong. Surface diffuse color and
   transparency, are completely separate phenomena, as you might
   observe it you look through a partially transparent balloon, for
   example. If we wanted to create a perfectly transparent glass ball,
   on the other hand, we would have to take care to set the diffuse
   surface color to black, to make sure that the surface in fact
   contributed NO color to the final image.

The rest) You're right that there was a lot in this problem set to
guess at, or to come up with on your own. What you did for angular
attenuation and eta-accumulation, for example, were fine.

--------------------------------------------