Create a two exact rectangles and choose one and punch in .5 in the corner radius from the tool bar. Now on the other use the Fillet docker and punch in .5. Center them over each other. When viewed in wirerframe and you will see a noticeable difference on the corners. Not enough to make a big deal but I would expect no difference.
Zoom @ 255793%. Like Eskimo, I find it interesting. Just striking up conversation.
Interesting indeed.
Wonder why there is any difference at all.In theory there shouldn't be.
I find this behavior of CorelDraw interesting, but not troubling.
It makes sense to me that somewhere, at some level, curves created using different methods are not going to be absolutely identical to each other. In the examples I've seen so far, that "deviation from perfection" is happening at a level less than 0.01 mm.
It also makes sense to me that having two very nearly, but not quite identical curves could cause problems for the "smart fill" when it's trying to find boundaries.
Suku said:True, its possible curves to have differences based on methods used to create. But when you intersect two objects you would expect the curves to match exactly.
I wouldn't expect the geometry resulting from a "trim" operation to necessarily create a curve exactly like the original.
In your example of using an ellipse (a circle, in that case) to trim a rectangle, it makes sense to me that the curved part of the new shape won't have exactly the same geometry as the ellipse did.
It's not history-based modeling. It's not "remembering" all of the details of the objects used to create it. Instead, it's creating new Bézier curves (with in some cases new nodes) that closely approximate those of the reference geometry.
Eskimo said: suku True, its possible curves to have differences based on methods used to create. But when you intersect two objects you would expect the curves to match exactly. I wouldn't expect the geometry resulting from a "trim" operation to necessarily create a curve exactly like the original. In your example of using an ellipse (a circle, in that case) to trim a rectangle, it makes sense to me that the curved part of the new shape won't have exactly the same geometry as the ellipse did. It's not history-based modeling. It's not "remembering" all of the details of the objects used to create it. Instead, it's creating new Bézier curves (with in some cases new nodes) that closely approximate those of the reference geometry.
suku True, its possible curves to have differences based on methods used to create. But when you intersect two objects you would expect the curves to match exactly.
I agree, Moreover, we have at least 3 situations:
Sometimes it's just a visual difference on screen, same if you see a thin border between two objects, but it's just visual, there's no difference and the printed result is fine. And sometimes, although the small difference exist, most printers couldn't reproduce it, and when you output there' not diference at all. Of course, it willbetter if we can see exactly the result on screen