I tried blurred reflections using averaged micronormals of 9 built-in 
patterns.  10 normals are averaged in each frame.  I used a bump size of 
0.1 for all patterns, so the differences in blurriness between frames 
are intrinsic to the patterns.

I used Alain's trick because without it, a render would occasionally get 
stuck.  I was puzzled because I only put one reflective object in the 
scene, but later it occurred to me that the lighting fixtures in my 
prefab render rig have reflective finishes.  Apparently, it doesn't take 
much to bog down a render.

The timing statistics are below.  The CPU is a quad core with dual 
registers.  Figures may not add due to rounding.  Radiosity times are 
omitted because I used two-pass radiosity: a radiosity file was created 
without normals, and this one file was loaded for all of the patterns. 
Parsing and bounding times were negligible.

Elapsed time (m:s)

   Pattern       Photons  Trace   Total
   -------       -------  -----   -----
   agate           0:34    3:30    4:04
   bozo            0:26    2:04    2:30
   bumps           0:26    2:43    3:10
   crackle         1:37    2:54    4:31
   dents           0:27    2:07    2:34
   facets coords   0:42    2:57    3:39
   granite         0:27    3:07    3:35
   spotted         0:27    2:01    2:28
   wrinkles        0:27    2:32    2:59

CPU time (m:s)

   Pattern       Photons  Trace   Total
   -------       -------  -----   -----
   agate           1:53   24:19   26:11
   bozo            1:37   14:48   16:25
   bumps           1:36   19:25   21:01
   crackle         3:15   20:48   24:04
   dents           1:37   15:11   16:48
   facets coords   2:05   21:23   23:27
   granite         1:44   22:13   23:57
   spotted         1:38   14:33   16:11
   wrinkles        1:42   18:04   19:47

Post a reply to this message

Attachments:
Download 'fuzzy_normal_types.jpg' (151 KB)

Preview of image 'fuzzy_normal_types.jpg'