| 
by software artist Scott Draves.  You may also follow me on google+ or twitter, buy art, or join me on facebook.
 | 
 Erik Reckase has implemented a Density Estimator for the FLAM3 algorithm.  This substantially reduces the spackle, dust, and noise too often currently visible.  Density Estimation works by making the
filter size inversely proportional to the hit density:  highly sampled areas need only a small filter, but
sparse areas need a large filter.    Adaptive Filtering for Progressive Monte Carlo Image Rendering by Frank Suykens and Yves D. Willems has the specifics.
Erik Reckase has implemented a Density Estimator for the FLAM3 algorithm.  This substantially reduces the spackle, dust, and noise too often currently visible.  Density Estimation works by making the
filter size inversely proportional to the hit density:  highly sampled areas need only a small filter, but
sparse areas need a large filter.    Adaptive Filtering for Progressive Monte Carlo Image Rendering by Frank Suykens and Yves D. Willems has the specifics.
Erik has a number of examples on his DeviantArt account, including this before and after comparison. Plus: this, this, and this.
The code will be released as soon as its cleaned up and the interfaces
are finalized.
Posted by spot at September  9, 2005 07:09 PM