Home

About

Advanced Search

Browse by Discipline

Scientific Societies

E-print Alerts

Add E-prints

E-print Network
FAQHELPSITE MAPCONTACT US


  Advanced Search  

 
Object-Space Variance Estimators John Amanatides
 

Summary: Object-Space Variance Estimators
John Amanatides
Dept. of Computer Science
York University
North York ON M3J 1P3
Canada
amana@cs.yorku.ca
ABSTRACT
This paper describes an exploration of anti-aliasing techniques for ray tracing that glean
variance information from object space rather than relying on samples to detect regions of alias-
ing.
Introduction
One of the biggest problems with ray tracing is that it is inherently a point-sampling process. Because of this anti-
aliasing is problematic. The usual solution is adaptive supersampling [1], and it is used to reduce both aliasing and
total computational cost. Unfortunately, adaptive point sampling is a post-sampling process and can be fooled.
Because the samples themselves are affected by aliasing they may not trigger an increased sampling rate in problem
regions.
What are needed are good estimators for variance, hopefully ones that are not as sensitive to aliasing. For example,
when one is near a silhouette or on a textured object aliasing is more likely. If one knows this one can increase the
sampling rates in these regions. One can imagine an advanced adaptive sampler that returns a variance estimator

  

Source: Amanatides, John - Department of Computer Science, York University (Toronto)

 

Collections: Computer Technologies and Information Sciences