Here's another of those not-quite-right things that have been around for 
ages:

When adaptively sampling a media interval, the sampling pointsare 
weighted as follows (example):


               sub-intervals good enought to
          not require maximum-depth oversampling
                |                         |
                v                         v
   |---|---|-------|---|---|---|---|---------------|---|---|---|---|
   1   1   3       2   1   2   1   5               3   1   2   1   1
                                 ----
                                  24

Note that...

(1) the weighting of samples taken at maximum-depth subsampling 
alternates between 2 and 1 (the points at the interval borders fit in 
this pattern as well, as they will also affect an adjacent media 
interval if there is one); and

(2) the weighting of samples at the border of less-precisely-sampled 
intervals is unbalanced.

I'd love to fix this as follows:

   |---|---|-------|---|---|---|---|---------------|---|---|---|---|
   1   2   3       3   2   2   2   5               5   2   2   2   1
                                 ----
                                  32

Anyone out there who thinks such a change could mess up existing scenes?

Post a reply to this message