Getting weird result with Real iteration count

[XSource]

Hi everyone. I'm working on a program to render the Mandelbrot set. I got it working for integer valued iteration counts, so I thought I'd give Real valued iteration counts a try. The way I calculate it is with

Code:

nu = n + 1 - log(log(|Z|))/log(2)

though I have tried other equations, but every time I get this result http://i.imgur.com/sRO6B.jpg (the white interior is intentional, it's the weird smoothing that's the problem). Has anyone seen anything like this before? It looks as though the smoothing is going in the wrong direction. That is, the gradient should be "facing in", if you can see what I mean in the pic.

[cKleinhuis]

try to use higher bailout values
david will know the answer, i am not a smooth coloring expert, thus havent checked your formula 

[XSource]

I did read about some equations based on very large bailout values, but I remember a few years ago seeing one that worked with a bailout of 2

[simon.snake]

Looks like one of my FractInt formula versions of the mandelbrot:

http://www.fractalforums.com/index.php?action=gallery;sa=view;id=11281