cropping/masking a bitmap

[Christine Farrelly]

Your right it appears to work sometimes - perhaps Xara can shed some light on this. I must admit I usually use a combination of the methods described here. Get the shape nearly right then chip away any stray bits zooming in to do fine bits. Then feathering the edgest to blend into new setting.

Christine

[Bob C.]

The closed shape remains.

I must admit I don't use this technique much but when working on Ross's hair challenge I carefully traced Ross's image using the Freehand tool and closed the shape (no fill) Made minor adjustments with the Pen tool (not the Shape Editor tool unless I really needed it)

Selected both (the bitmap and my closed trace) then from the "arrange" menu picked "combine shapes/intersect shapes" and the background disappears.....every time. Well I tried 3 times [img]/infopop/emoticons/icon_wink.gif[/img]

[Mickie]

Alan, I didn't keep the bird. It's a xara photo from the clipart gallery. I've got the feeling that you and Christine have come up with the general "rule" but there must be some sort of "i before e except after c rule" maybe it doesn't recognize nodes added or something.

Bob, your method works, but I have found I prefer snipping away background with the freehand tool until I'm close to the shape I want. Then I work with the nodes to get it right.

Mickie

[Egg Bramhill]

This one is strange. I've been playing about with it since the original posting. I believe it is the node count that matters, but it seems to follow some rules.
If the image is sliced, with equal number of nodes on each side, delete the section that does NOT have the starting point within it. (Even if this is the larger shape)
If the image is sliced across the starting point, then the node count must be greater than 2 on the non starting point section before the section with the starting point is deleted. (Remebering to add the new added nodes to both sides in the count.

This sounds complicated, but I'll try to do a mini tut. (Not that its that important but it would be nice to know)
Egg

[Ross Macintosh]

I'm glad to see some energy going into understanding this one.

It just struck me that most of us are talking about this in the context of editing bitmap-filled shapes to facilitate making composite images. I thought I should also perhaps remind that these techniques can be part of your general arsenal of shape editing approaches - if they aren't already. Isn't it great how xara's tools are so flexible! (The cad program 'intergraph' used to be famous for having a tool for everything. I much prefer the xara approach of having a few very flexible tools. I'm too stupid to learn 237 tools [img]/infopop/emoticons/icon_biggrin.gif[/img] ).

Regards, Ross

<a href=http://www.designstop.com/>DesignStop.Com</a>

[Egg Bramhill]

Yes, it struck me that we were all talking about bitmaps, but any shape will do just as well. I've posted a long explanation that gets no nearer the truth but it may be of interest in this thread. By the way even bitmaps have a start point in Xara so this would explain Mickie's finding, as he was cutting out the start point corner.
Re Bob C's point: The upper shape will always slice the lower shape Bob. Xara always works this way, and knowing it makes slicing 2 objects far more predictable.
slicing images using the freehand tool
Egg

[Ed Nadie]

Egg,

I looked at your slicing exploration, and can suggest that straight and curved lines need to be differentiated. It's easy to verify that curved lines have more 'weight' than straight ones, and would help explain the seemingly inconsistent results.

These variables can probably be ruled out as factors:
» size of enclosed area
» relative density of nodes (or length of line segments)
» cartesian orientation (eg. top versus bottom)
» direction of the line
» chronology of changes (revisions, most recent operation, etc.)
» type of slicing line (doesn't matter if it's straight, wavy or loopy -- it will attach to whichever side wins the slicing derby.)
» type of tool used to create the shape.

What is left?
The mathematical complexity of the line as defined by the number of nodes and number of bezier tangent arms. That would probably be the simplest and surest way to predict the most important division of a shape.

It's clear that it takes two extra nodes (without handles) to overcome the advantage of a shape's starting point -- so the starting point would get a bonus weight somewhere between one and two nodes, call it 1.5 (it gets one point for just being a node, so that's total of 2.5; and more if it's got control handles).

I suspect bezier tangent arms probably have a similar weighting. (By comparing your examples nine and ten, you can estimate that each tangent line might be weighted about 0.7.) Nodes with two tangents likely have twice the additional weight of nodes with just one tangent line. Possibly the number and type of line segments (straight vs curved) is considered in addition to, or instead of the nodes. There may be other considerations, when shapes have donut holes, etc.

I like to draw with the freehand tool with lots of loops, and also use the 'join shapes' command a lot, so I end up with a single convoluted shape, and often ninety percent of my drawing will disappear when I try to redraw a small corner. I end up either deleting nodes individually, or draw a shape over a complex area I want to redo, and 'subtract shapes'. I hope someone from Xara will respond, because I've also wondered about this.

-- Ed Nadie

[Mickie]

Attached is another experiment. I'm assuming the start point for a bitmap is the lower left. If I slice a bitmap in half, then I should always retain the lower part. My example uses curved lines, one slice beginning in upper left to lower right, the other lower left to upper right...and I have two different results.

All I can say is thank goodness for the undo command!

Egg, by the way I'm a she. In the U.S. most nicknames ending in ie are female, but not all...as we are proving there are exceptions to every rule. [img]/infopop/emoticons/icon_wink.gif[/img]

Mickie

[Egg Bramhill]

Ed,
You're probably correct in your assumptions. The more you mess around with the slicing the more complex it becomes. It would be interesting to know how the XaraX algorithim decides these factors, but what practicle use would it be, I'm not sure! Could you imagine trying to work out some calculus to decide what the end result would be... far better to try and use the undo icon. But it's something of a puzzle which is interesting to try to unravel.

Mickie,
Sorry for the gender confusion!!!! Re the example, the lower left is not the start point of the bitmap, but the top left. If you select the bitmap with the selector tool you get the 8 handles and the start point indicator. See attachment. It's not a well documented feature in XaraX but its handy when your trying to find start points within several objects.

Egg

[Ed Nadie]

Mickie,

The attachment refers to your second experiment, and shows that the
starting point has a total weight of somewhere between two and three
regular nodes. The cyan examples show that curves in a slicing line don't
make any difference. Curved lines are only calculated as an indication of
complexity -- and therefore likely importance -- of the shape you're
slicing (like your first experiment).

With a rectangular bitmap, you can easily predict which side will remain,
if you know the starting point, but in a shape with curved edges, even
knowing the exact algorithm wouldn't take the place of the undo button.

-- Ed Nadie