weird dip in profile display

[gr5]

seen in v8.3.2 but I think all versions probably do this.

Minor bug.

Dale had an extremely similar issue (that I closed) that was in an older version but that got fixed I think but I think this issue is different. Maybe.

If you have a wavefront that is an odd number of pixels (601 pixels for sure works for the bug) then the profile, when perfectly horizontal (at 0 degrees angle), will have one zero value at the end and the profile sharply dips up or down on the last pixel.

It only happens at 0 rotation where the above mentioned for loop is getting a floating point off-by-one at the end of the loop.

[githubdoe]

Yes I have seen that. Sounds like an interaction with rounding to an int and spilling over into the mask region.

[gr5]

I don't think it's the masking. I think it's that the profile loop goes past the edge of the wavefront. But I don't know for sure. I looked at it very carefully in 3D view and there is no "bad" point on the right edge. I'll look at this some day as it's easily repeatable.

[githubdoe]

I find it only happens when the wave front is created the first time. If I then rotate the diameter it goes away and it can never be seen again. Until you modify it again. I think for some reason the first draw goes on pixel too far to the right. Can't figure out why. If it only happens on odd widths I sure think it is a round up error.

Also I just saw it on a diagonal diameter. Not a horizontal one. So it can happen there as well.

[gr5]

Yeah it only happens on odd widths.

I can rotate around and get back to the error but it's not easy.