Neue Testversion 18.40b3

[photoken]

Yes, I can reproduce that behaviour - at 50% the triangle's stroke is broken like in your example - but only when I set the line join to either rounded or flattened - with sharp line joins it works as expected.

Thanks. I hope we've given the developers enough info to track down the problem....

Fixed in 18.40b4. Thanks!

[Martin Huber]

But the treatment of the antialiasing of masks was a bit of a surprise to me. It seems like a bug that "antialias always" is ignored for adjustment layers and layer masks until the layers are fixed.

This should work in the next beta.

By the way: What about replacing "always" by a list of interpolation methods?

The list would be quite short: "Bilinear" and "Bicubic". Other interpolation methods are not supported in this situation. And is there really a reason for "Bilinear"?

Martin

[Hoogo]

The list would be quite short: "Bilinear" and "Bicubic". Other interpolation methods are not supported in this situation. And is there really a reason for "Bilinear"?

Masks with lower resolution are somewhat tricky, as their pixels do not match the pixels of the picture. I think the best use for this might be gradients, and these look better with a smoother interpolation. Not sure if this makes really a difference...

[bkh]

I think the best use for this might be gradients, and these look better with a smoother interpolation. Not sure if this makes really a difference...

For smooth transitions such as gradients, the difference between bilinear and bicubic should be invisible (except that the latter takes a bit more time to compute).

Cheers

Burkhard.