Ok so I interpret that we have a consensus of the following: 1) Use a "multichannel" named argument that defaults to True, rather than try some fancy guesswork based on the input data dimension. 2) Make 2D a special case of 3D to keep codebase simple, at a performance cost relative to a pure 2D implementation. I'll go ahead with this implementation unless someone objects... Also, I intend to assume isotropic 3D to begin with, and add support for non-isotropic data in a subsequent PR... Does that make sense? Or do you prefer I do it all at once? On Mon, Mar 11, 2013 at 6:52 PM, Stéfan van der Walt <stefan@sun.ac.za>wrote:

On Mon, Mar 11, 2013 at 9:39 AM, Ronnie Ghose <ronnie.ghose@gmail.com> wrote:

...

... ehhh... I would prefer 2d rgb to be the default in general as I think that is the generl use case? . This calls for a discussion imho.

Yes, I think for this algorithm you're right.

Stéfan

-- You received this message because you are subscribed to the Google Groups "scikit-image" group. To unsubscribe from this group and stop receiving emails from it, send an email to scikit-image+unsubscribe@googlegroups.com. For more options, visit https://groups.google.com/groups/opt_out.