[vtk-developers] New vector modes for vtkScalarsToColors

[David Gobbi]

On Thu, Apr 28, 2011 at 9:36 AM, Nicolas Rannou
<Nicolas_Rannou at hms.harvard.edu> wrote:
> Hi David,
>
> Sorry for the late reply... We are now ready to step into this problem.
>
> We have N images corresponding to N channels.

I thought you had one image with N channels (i.e. one N-component
image), that is the only situation where my proposal is useful.  For N
images each with its own channel, you should do something entirely
different:


image_1 -> vtkImageMapToColors (with LUT #1) -> vtkImageWeightedSum -> display
image_2 -> vtkImageMapToColors (with LUT #2) ---/^



> Is it what you suggested or did I misunderstand you?

I proposed that vtkScalarsToColors could simultaneously apply
different tables to different channels _in a single image_ , and then
combine the results (e.g. by summing the colors resulting from each
channel and clamping to a max value of 255).  So a pipeline using this
proposed capability would include only vtkScalarsToColors and
vtkImageMapToColors.  No other classes would be needed.  (But note
that this doesn't fit your situation, since you have N images, rather
than one image with N channels).

The reasons for adding this ability to vtkScalarsToColors rather than
making a new lookup-table class are (1) adding it to
vtkScalarsToColors would allow this multi-capability to be used with
both lookup tables and with color-transfer functions and (2)
vtkLookupTable does not subclass as well as you might think, because
of the way it is implemented its subclasses can only change its
behavior in trivial ways.

 - David