Gradient bugs in direction and length

[LWChris]

When I make a linear gradient from the bottom left corner roughly in 45° angle, then I am able to bug the gradient into one that only runs though a small amount of the actual gradient range that should be applied.

 

Steps to reproduce:

 

1) Open a new image

2) Select the gradient tool

3) Draw a gradient from the left bottom in about a 45° degree angle (to the top right)

4) When you slowly drag around the handle close to the image border, at some spot you can cause the visual gradient to bug into a horizontal direction and twice the length.

 

Reproducible: always (10/10 tries; varying colours [red green, blue orange, unchanged black white], different canvas sizes [square, almost square, rectangle, unchanged 800x600])

Note that only the visual appearance and the outcome is affected, while the displayed toolbar values and handles stay at their correct positions.

 

The appended screenshot should clarify what I am seeing.

 

Windows 10 Insider Preview Version 1511, paint.net 4.0.9 (I know the screenshot says 4.0.8 but I was already able to reproduce it in 4.0.9, too)

[Ego Eram Reputo]

Confirmed. When dragging out a diagonal gradient top right to bottom left there is a pixel boundary which messes with the gradient.

 

The boundary seems to be a single column of pixels close to the left hand side. Repros with both rectangular and square images. The reverse direction shows the same bug (bottom left to top right). The other diagonal is Ok.

 

Hypothesis: it appears to be the same number of pixels from the left hand side as the number of pixels down from the top that the gradient started.

 

Repro:

 

 

1. Drag a linear gradient from A to B. Note the number of pixels (x) the gradient starts from down from the top edge.

 

2. When you get to X pixels from the left hand edge the gradient changes color as you horizontally traverse the red column of pixels.

Here's what the bug looks like....

[Rick Brewster]

Can you give me the exact pixel coordinates and lengths and stuff, from the status bar?

[Eli]

Hello Rick, It is hard to find the exact place but finally found it.

 

 

[Rick Brewster]

Okay I managed to repro this

[Rick Brewster]

I figured this out and will have a fix for the next update

 

It was a typo. If the ending X coordinate is equal to the starting Y coordinate, then you'll see this behavior.

[Ego Eram Reputo]

It was a typo. If the ending X coordinate is equal to the starting Y coordinate, then you'll see this behavior.

That's what I was trying to say