I've always wondered that. I'm not familiar with the code but this sounds more like a limitation of the method.

Putting the factor value to 0.0 will result a fully unblurred results though so it could be checked but I'd imagine this is not on the top of the priority list...

Also fast blur ignores the factor value.. if this is just a limitation of the method it needs clear UI indication :)

Well if it's really the limitation, I'd suggest making -1 operation on both blur radius and limiting the value to be always at least 1.

The fast blur ignoring the factor (or size as it's now called) is new to me... But seems to happen in 2.49b too. Hide the size node input?

Lets see what the devs think about this but this has never been a biggie for me.

fixed r32419.