Image Synthesis Class

Homework 2:

Sample the function: L(x,y) = 0.5 * ( 1 + sin( 0.01*x² + 0.01*y² ) ) with various sampling functions. Compare the results to the box filtered image from Homework 2 to compute absolute and relative error as well as the average errors.

NOTE: All images are gamma corrected with gamma 2 (for relatively good viewing on all computers).

NOTE: All error images (and average errors) were computed using the values sampled directly from the function L(x,y). Errors were not computed from the PGM files (which round to integer values).

The "ideal" image
(L(x,y) sampled 10,000 per pixel with a Box Filter)

64 Samples Per Pixel:

	Image	Absolute Error Image	Relative Error Image
Regular Sampling		Average Error: 0.0025	Average Error: 0.0060
Random Sampling		Average Error: 0.0282	Average Error: 0.0654
Jittered Sampling		Average Error: 0.0055	Average Error: 0.0126
Multi-jittered Sampling		Average Error: 0.0039	Average Error: 0.0089
Hammersley Sampling		Average Error: 0.0092	Average Error: 0.0228

16 Samples Per Pixel:

	Image	Absolute Error Image	Relative Error Image
Regular Sampling		Average Error: 0.0055	Average Error: 0.0138
Random Sampling		Average Error: 0.0564	Average Error: 0.1313
Jittered Sampling		Average Error: 0.0196	Average Error: 0.0457
Multi-jittered Sampling		Average Error: 0.0136	Average Error: 0.0307
Hammersley Sampling		Average Error: 0.0245	Average Error: 0.0608

4 Samples Per Pixel:

	Image	Absolute Error Image	Relative Error Image
Regular Sampling		Average Error: 0.0224	Average Error: 0.0566
Random Sampling		Average Error: 0.1139	Average Error: 0.2641
Jittered Sampling		Average Error: 0.0731	Average Error: 0.1711
Multi-jittered Sampling		Average Error: 0.0607	Average Error: 0.1394
Hammersley Sampling		Average Error: 0.0566	Average Error: 0.1405

Last Modified: Tuesday, Sept 11, 2001

Chris Wyman (wyman@cs.utah.edu)