Why these rounded corners?

[flagpole]

I don't think this is a bug. I think it's expected behaviour at least. It happens because it's trying to expand the selection by a certain amount that is the radius of the curve.

I can't imagine that it would ever be any different.

[JoeB]

I don't think this is a bug. I think it's expected behaviour at least. It happens because it's trying to expand the selection by a certain amount that is the radius of the curve.

I can't imagine that it would ever be any different.

I suppose it depends on your perspective. In other words, the technical aspects aren't of concern to users. Expectations are what counts. And when a user starts with a square cornered rectangle and wishes to expand that rectangle the expectation is that the result will be a square cornered, larger rectangle, the same as reducing the rectangle is expected to produce a square cornered, smaller rectangle - which it does. So it's definitely a bug in the sense that it doesn't produce the result that the user has every reasonable expectation that it will produce. And that doesn't have to be the case as shown by the workaround which should actually be built into the Expand command. Nobody expects that they will not get a square cornered rectangle when they expand a square cornered rectangle. If they wanted a rounded rectangle they would naturally use the Rounded Rectangle tool which, when expanded, would still produce a rounded rectangle. I suspect people would be quite put out if they started with a rounded rectangle and, when expanded, it suddenly turned into a square cornered rectangle for some behind the scenes technical reason.

[flagpole]

I don't think this is a bug. I think it's expected behaviour at least. It happens because it's trying to expand the selection by a certain amount that is the radius of the curve.

I can't imagine that it would ever be any different.

I suppose it depends on your perspective. In other words, the technical aspects aren't of concern to users. Expectations are what counts. And when a user starts with a square cornered rectangle and wishes to expand that rectangle the expectation is that the result will be a square cornered, larger rectangle, the same as reducing the rectangle is expected to produce a square cornered, smaller rectangle - which it does. So it's definitely a bug in the sense that it doesn't produce the result that the user has every reasonable expectation that it will produce. And that doesn't have to be the case as shown by the workaround which should actually be built into the Expand command. Nobody expects that they will not get a square cornered rectangle when they expand a square cornered rectangle. If they wanted a rounded rectangle they would naturally use the Rounded Rectangle tool which, when expanded, would still produce a rounded rectangle. I suspect people would be quite put out if they started with a rounded rectangle and, when expanded, it suddenly turned into a square cornered rectangle for some behind the scenes technical reason.

Interesting.

The thought would never have occurred to me that I would get square corners when expanding a selection. What would you expect when expanding the selection around a single pixel?

I agree the solution is to give a simple check box in the expand dialogue.

It could be worse. I used to use a program that would expand a single pixel into a diamond shape.

[JoeB]

Interesting. The thought would never have occurred to me that I would get square corners when expanding a selection.

Users of PSP don't ruminate about the technical aspects of how or why a tool works. Virtually everybody who has created a selection expects that selection to retain it's original shape whether expanded or contracted. You can't seriously believe that a user, planning on expanding his/her rectangular selection to create a border selection inside of it and fill it, really wants to get a rounded border on the outside of that frame - which, incidentally, will be jagged, not smooth - or actuallly EXPECTS to get a rounded border on the outside of that frame any more than they expect to get a different shape when they reduce the size of the selection. Of course they don't, which is why since the Jasc days there have been posts about it and why the workaround was suggested many, many years ago.

What would you expect when expanding the selection around a single pixel?/\

I have been doing graphic work for decades and have yet to see a need to expand a selection around a single pixel, and I suspect that applies to 99.9% of PSP users.

I agree the solution is to give a simple check box in the expand dialogue.

I can't imagine what a checkbox would be used for. I can't imagine why a user would create a square cornered rectangle and, when deciding to expand it, they would decide to check an option about whether or not they want the rectangle to retain its square corners or have rounded corners. If they had wanted rounded corners they would have selected the Rounded Rectangle tool in the first place.

The bottom line is that in all these years I have never heard one user post that they expect - or desire - rounded corners when expanding a square cornered rectangular selection, but many posts with people expressing dismay and concern when they do get that unwanted and unexpected result. Thus, as I said, the workaround that was probably posted before Corel even owned PSP. I know it was an issue for many of us back in the Jasc days when there weren't these web forums but only bulletin board news groups requiring newsgroup readers. Ahhh, for the good old days!

[flagpole]

Interesting. The thought would never have occurred to me that I would get square corners when expanding a selection.

Users of PSP don't ruminate about the technical aspects of how or why a tool works. Virtually everybody who has created a selection expects that selection to retain it's original shape whether expanded or contracted. You can't seriously believe that a user, planning on expanding his/her rectangular selection to create a border selection inside of it and fill it, really wants to get a rounded border on the outside of that frame or actuallly expects to get a rounded border on the outside of that frame any more than they expect to get a different shape when they reduce the size of the selection. Of course they don't, which is why since the Jasc days there have been posts about it and why the workaround was suggested many, many years ago.

What would you expect when expanding the selection around a single pixel?/\

I have been doing graphic work for decades and have yet to see a need to expand a selection around a single pixel, and I suspect that applies to 99.9% of PSP users.

I agree the solution is to give a simple check box in the expand dialogue.

I can't imagine what a checkbox would be used for. I can't imagine why a user would create a square cornered rectangle and, when deciding to expand it, they would decide to check an option about whether or not they want the rectangle to retain its square corners or have rounded corners. If they had wanted rounded corners they would have selected the Rounded Rectangle tool in the first place.

The bottom line is that in all these years I have never heard one user post that they expect - or desire - rounded corners when expanding a square cornered rectangular selection, but many posts with people expressing dismay and concern when they do get that unwanted and unexpected result. Thus, as I said, the workaround that was probably posted before Corel even owned PSP. I know it was an issue for many of us back in the Jasc days when there weren't these web forums but only bulletin board news groups requiring newsgroup readers. Ahhh, for the good old days!

I seem to have offended you.

I wasn't suggesting that there would be a need to expand a single pixel. I was using it to illustrate a point. It was a rhetorical question. That I would expect expanding a single pixel to yield a circle. Not a square.

It's useful when working with text.

I don't know how you would know what users who aren't you would want or expect. What goes through their minds. Or what they ruminate on.

But speaking for this user, I think it works correctly.

[LeviFiction]

I think the confusion people probably experience has to do with the word choice. The function seems to be based on the difference of "Expand" vs "Scale". When we use "expand" some are expecting the selection to simply resize itself, not balloon out and this would cause frustration. Both are perfectly valid methods with their own unique benefits and use cases, expanding vs scaling. Though scaling would result in a softer edge as it's inventing/averaging pixels.

A scaling routine in a script for PSP would probably work something like this
VerifySelection
EditSelection
PickTool scale - recenter
Cancel Edit Selection

Here is a demonstration image: You have the original selection, what the selection looks like after being expanded by 10 pixels, and what the selection looks like after being resized by 10 pixels.

Is the "resized" example what you'd expect from an expand JoeB? Or would you expect a mixture between Resized and Expanded? You'll notice that expansion expands all areas of the selection, taking up more actual selected pace. Resizing also takes up more space but never into itself. It's always outward, the holes resizing to maintain their consistency with the rest of the selection. This allows us to maintain the sharp angles. Uncertain how a mixture of the two would work. Decreasing holes while making sure each whole is scaled uniformly?

[flagpole]

So to my mind, what i meant anyway, was that you can do stuff like this...