rotated path essue

[denzjos]

My drawing included a circle path and I have rotated the path assy. I've made intemediate circles paths (Ofnuts plugin) from this circle and an enlarged one (same center). I detemined that the new circles are a little out of the original center. I've searched what was the cause and saw, when I enlarge a rotated circle, that the centerpoint is not the same. I think gimp is calculating with pixel units and shift the middle point to the cross line of the pixels in the direct environment. I just want to mention this essue. On the other hand it is possible to move paths by part of a pixel by using the rotation tool : just play with the rotating point. Very small essue to discover, but I'm glad I found it.    

   
   
   

[Krikor]

What I could see is that after rotating the circles their diameters were changed.
The original circle is 200x200.
Made two copies and rotated both at an angle of 20.57, the dimensions of both were changed to 202x201.

   


By centering on the screen, the enlarged circles had a different center.

   

[denzjos]

@Krikor, I also noticed that. That's probably because if rotated the circle is no more on the crossing line of pixels. I think that the values from the coordinates on the new position are rounded to an even value and this is reflected on the circle. I could be wrong.

[Krikor]

@denzjos,

I thought the problem could originate if the center of rotation was not an integer (non-decimal) point (x,y).

But I did some tests, creating rotation points in integer coordinates and even then the problem was repeated.

In the image below, the original circle (100x100) after being rotated using the Rotate Tool  , was changed to 101x101.

   

I did another test, this time using the plugin (Duplicate and Rotate a Path...), and as the image below illustrates, there was also a change in the dimensions of the circles after rotation in 100% of the created and rotated copies.


   

[Ofnuts]

A path circle isn't a perfect circle, it is an approximation.Since it is not perfect, the diameter isn't constant and under some angles it can be bigger or smaller than the ideal diameter, so you can indeed see an increase of diameter when you rotate it. In paths constructed with ofn-path-to-shape, points at 45° intervals (starting from the segment that was used to construct the circle) are exactly on the circle.

[denzjos]

(01-21-2023, 02:18 PM)Ofnuts Wrote: A path circle isn't a perfect circle, it is an approximation.Since it is not perfect, the diameter isn't constant and under some angles it can be bigger or smaller than the ideal diameter, so you can indeed see an increase of diameter when you rotate it. In paths constructed with ofn-path-to-shape, points at 45° intervals (starting from the segment that was used to construct the circle) are exactly on the circle.

Ofnuts, thank you for this explanation