ZDefocus plugin

[devernay]

inputs are:

• a depth or disparity map
• an image

parameters are:

• boolean: is it depth or disparity?
• boolean: is the focus distance set using a point interact or a scalar value?
• double2D: point interact (disabled if focus is set using a value)
• double: focus distance
• depth of focus
• number of focus slices
• apperture shape (Gaussian apperture uses a fast blur)

Algorithm: create masks for each focus slice, blur premultiplied image+mask with the right blur radius (using convolve, for a nice bokeh effect), and merge over successively each blurred image from far to near focus slices, to handle occlusions correctly.

In draft mode, always use Gaussian blur