Artur Rataj wrote:

> Hello, can anyone point me to some reference of the algorithm?

I don't know the exact source that the programmers were referring
to, but if you google for "fast gaussian blur" you will find 
plenty of references.

> Why the translation radius -> std dev is so strange?

My understanding is that it is basically a kludge.  Simply
using the specified radius as the standard deviation of the
Gaussian would be the "natural" thing to do, but the result
looks like it smears things more than the user expects.  The
formula used by GIMP, I think, is just an attempt to set a
radius that makes a sharp dot look perceptually like a fuzzy
circle with something like the specified radius.

  -- Bill

______________ ______________ ______________ ______________
Sent via the CNPRC Email system at

Gimp-developer mailing list

Reply via email to