Problem with mask

[Wilhelm Sanke]

On Thu Oct 16, 2008, James Hurley jhurley0305 at sbcglobal.net wrote:

> It worked after a fashion but is rough around the edges. In  
> particular when masking to an ellipse, the results is flat on the  
> right edge and the bottom edge. I do this by running over the image  
> and masking all pixels that are not within the ellipse--using the  
> "within" function. That should work.
>
> But it doesn't quite. There must be something  about images I don't  
> understand. I would appreciate any thoughts.
>
> To see the effect, run this in the msg box:
>
> go url "http://home.infostations.net/jhurley/MaskingToEllipse.rev"
>
> Jim Hurley


Hi Jim,

as usual, somewhat late in joining this thread.

I tried to download your stack several times, but all I get is "No such 
card" in the msg, so I cannot look at what may have caused those flat edges.

We had a related discussion in December 2007 ("Onion Skinning", 
"Caricature Challenge") where I had described the routines I apply to 
mask variable parts of an image - possibly resize (shrink and enlarge) 
and/or rotate the masked image - and which is then pasted into another 
part of the same or another image. But the newly masked and cropped 
image could also be used as a separate new image.

Such routines will be part of my upcoming stack "Photo Patchworks" which 
contains 6 scalable and draggable selection graphics for masking, namely 
rect, oval, rhomb, star, triangle-up, triangle-down, and semi-circle. 
These selection graphics are accompanied (invisibly) by corresponding 
and partly transparent png-images, which have different "fringes" of 
transparency among which you can choose for masking.

Taking a snapshot is not part of the process.

I repeat part of my post of Dec, 18, 2007 ("Caricature Challenge") here, 
and would still like to have a look at your stack. Could you upload it 
again?:

> In my reply of Nov 29 to thread "Any suggestions on how to "onion
> skinning"?" I had mentioned some new features of my upcoming release of
> the "Imagedata Toolkit" - among them
>
> "copying - and enlarging or shrinking - and pasting oval or rectangular
> portions of an image into the same or another image with variable fringe
> and/or overall blending into the basic image)."
>
> I applied this to a photo of a not unknown public figure
>
> (snipped here, as a number of list members felt offended by the 
> caricature)
>
> To achieve this I used a number of steps, in which about two of Ken's
> five steps are applied, but most of which follow a  different sequence
> and are of another  nature.
>
> The basis for the "copying - and enlarging or shrinking - and pasting
> oval or rectangular portions of an image"  are rects, ovals, and
> polygons in the form of images with transparent fringes of different 
> sizes.
>
> I use a rectangular or oval graphic as a selection tool that can be
> resized and dragged  across the photo to select a portion of it (This
> selection tool is similar to that used in my "seamless tiles 2" stack).
> The underlying image - the partially tranparent basic template - is then
> adjusted in location  and size  to the selected portion of  the photo.
> At the same time an already existing empty image "newimage" is  also set
> to  the selected portion of the image.
> The imagedata of  the selected portion  of the photo are then read  into
> the "newimage", and the alphadata of the template are applied to the
> "newimage".
>
> You can then enlarge or shrink  "newimage" or leave  it as it  is and
> drag it to any place  on the photo. You can  also flip "newimage".
>
> Next step is to integrate the image- and alphadata of "newimage" into
> the photo, either once or several times at different places. This
> integration can be applied with variable levels of blending,  in
> addition to the transparent fringe of the template that ensures optimal
> blending into the photo.
>
> If an enlarged portion  of  "newimage" should have been dragged to a
> position where it overlaps the rect of the photo, the overlapping parts
> of "newimage" are then cut off in the process of integration.-


Best regards,

Wilhelm Sanke

<http://www.sanke.org/MetaMedia>