|
![](/i/fill.gif) |
Glen Berry <7no### [at] ezwv com> wrote...
> My original idea was to keep track of the individual image pixel
> values in floating point format and only apply effects to pixels that
> *exceeded* 1.0 in value. The actual threshold would be changeable for
> example, to something like 3, 12, or whatever you want. This way,
> perfect white surfaces which might have a value of 1.0 would not be
> blurred, but light sources which might exceed a value of 1.0 by a
> significant amount *would* be blurred.
>
> In fact, this threshold could be applied to things other than the
> current blur feature. Lens flare, comma distortion, and "cross" or
> star-shaped highlights come to mind as possible candidates for effects
> that would depend on a pixel brightness threshold.
>
> Does anyone else like the idea?
I do. The functionality to do this (in terms of using the original,
unclipped float values) is already part of the post-processing framework.
If anybody want's to develop more post-processing filters and would like
some advice, let me know. Hopefully this weekend I will have some time to
write up a little tutorial on writing new filters.
-Nathan
Post a reply to this message
|
![](/i/fill.gif) |