Selection Handles Problems

[SearedIce]

I have the weird feeling that this has already been discussed, but I can't find any posts about it.

Here goes:

I make a very small selection (say, 2x2 pixels).

I zoom in so that I can see the selection well.

I activate either the "Move Selected Pixels" or "Move Selection" tool.

Now, when I try to grab handles for resizing, I have a hard time grabbing the correct handles. Clicking and holding DIRECTLY on some handles at this small selection size will not always grab the correct handle.

From what I know about programming, it seems that the current code is handling (yeah, bad pun) handle selection based on where the cursor is on the image, not where the cursor is on the computer screen. So, the region around each handle in which clicking will select that handle overlaps with the same regions for other handles, resulting in unpredictable handle selection behaviour for very small selection sizes.

If I'm correct, I might suggest changing this so that it determines which handle one winds up selecting based on the cursor's screen location rather than its image location.

Thanks for reading,

John

[Rick Brewster]

Yeah I think this was brought up very very recently in fact. In fact, it's in the post right below this one ("Move small selection").

The technical explanation is that, like you said, the hit-testing is performed based on where the cursor is "on the image". We don't currently have a consistent system for keeping tracking of floating point coordinates in document space. Anyway, I'm planning to fix it for 3.0, it catches me all the time when trying to resize small selections.

[SearedIce]

Yeah, that post is what prompted me to write this. I just didn't want to steal the thread. I was thinking that it was discussed some time in the past.

I'm glad to hear that a solution is in the works. I didn't think about a floating point coordinate system...that's interesting. That idea reminds me of another thing I thought I would bring up sometime:

In many of Photoshop's features, things can be done in fractions of pixels. For instance, you can do a .5 pixel gaussian blur. Will this ever be implemented? These sorts of fractional pixel inputs would make small-image editing much easier (or at least, much nicer). For example, if you were doing something with a 50x50 image and for some reason a 1 pixel blur were perfect for what you wanted, you would not be able to get your "perfect" blur for a version of the image that is 25x25 or really anything that is smaller than the original.

I know this could be a pain to code for something like a gaussian blur (I've read about the algorithms), but it's worth a look.