Kevin Harris
CS6620
Programming assignment #3
02Feb1999
The following images were generated with the camera at location
<0,0,-1>, looking in direction <1,1,0>, with vertical vield
of view at 90 degrees, horizontal field of view at 146 degrees. The
plane is sitting in the x-y plane (z=0).
The samples were jittered and triangle filtered across the virtual
screen, and a single random point was chosen on the disc for each
sample on the screen.
Note: When I refer to Fxx below, it would be similar to the F-stops
on a camera lens, where the radius of the sampling disc at Fxx is
focal_distance/(2*xx). I have been told that I need to say that the
F-stop is the aperture size (the size of the opening decreases as
the F number increases).
The images in these tables took a combined total of 5 hours, 9
minutes on a Pentium 166 to generate.
Just for humor, I calculated there to be 279,552,000 samples taken
in the lower 3 tables (that means that the Pentium was making a
little over 15 thousand samples per second). Running some benchmark
traces on one of the ultra60s had about 52 thousand samples per
second.
1 sample per pixel
| focal length | F8 | F32 | F128 | F512 |
|---|---|---|---|---|
| Focal length 1 |
|
|
|
|
| Focal length 4 |
|
|
|
|
| Focal length 16 |
|
|
|
|
| Focal length 64 |
|
|
|
|
16 samples per pixel
| focal length | F8 | F32 | F128 | F512 |
|---|---|---|---|---|
| Focal length 1 |
|
|
|
|
| Focal length 4 |
|
|
|
|
| Focal length 16 |
|
|
|
|
| Focal length 64 |
|
|
|
|
256 samples per pixel
| focal length | F8 | F32 | F128 | F512 |
|---|---|---|---|---|
| Focal length 1 |
|
|
|
|
| Focal length 4 |
|
|
|
|
| Focal length 16 |
|
|
|
|
| Focal length 64 |
|
|
|
|
Since I had some complaints about my field of view, I generated
another image, camera at <0,0,-8>, looking in the direction of
<8,8,0>, focal length 16, aperture F8, 512 samples per pixel,
size 640x480. It was created on an ultra60, and took 42 minutes 55
seconds to generate (61,082 samples per second).
