fuzzy selections

[Colin Walls]

This is something I have been trying to figure out and I have to ask the experts out there for some elucidation.

Broadly, the question is: how does the conversion from a mask to a selection work?

For example, using QuickMask [but this would be the same with layer masks], I can put a gradient across the image. When I swap to selection mode, I get a line of "marching ants" down the picture, which is slightly "jiggly". If I apply any effects [say desaturation], I get the progressive effect across the image, just like I'd get with a gradient mask.

So, does the mask get converted to a selection with feathering?

If so, something more complex than a simple gradient may require a selection with different feathering in different places. Is that possible?

I have arrived at this query having concluded that I much prefer using masks to selections. I'd just like to understand the details.

[Eric Purkalitis]

This is a slightly different topic, but I think it'll answer you're question.

Mark Reynolds "Loading Composite as Selection" 8/29/03 7:15am </cgi-bin/webx?13@@.2ccd2437/7>

[YrbkMgr]

Well, it all made sense until I read:

If so, something more complex than a simple gradient may require a selection
with different feathering in different places. Is that possible?




Can you rephrase what you're asking. Up to that point, I would've answered:

So, does the mask get converted to a selection with feathering?




as a "yes". But I suspect you mean something other than what I think. So if you could clarify it a bit, maybe with an example, it may help.

Peace,
Tony

[Colin Walls]

Thank you Mark. You have clarified the matter very well.

I had seen the mapping between feathering and a blur on a channel. The thing I couldn't understand was the fact that I could apply an uneven blur on a channel, but couldn't do uneven feathering when selecting.

The really confusing thing is the marching ants not being the boundary of a selection at all, but marking the mean [median?] point.

So using selection tools is fine, but working on a mask and viewing in that way gives much better control. Now I realise why I like masks so much and could never go back to PSE.

Again, thanks for the elucidation.

[YrbkMgr]

Unless a selection is very straightforwad, like with a marquee tool, I always check it by creating a temporary alpha channel and look at it there. Say you use the magic wand. The marching ants are almost never accurate so an alpha channel allows me to tell right away how much of what I wanted, I got.

[Colin Walls]

Tony

That's a good point. I guess the Magic Wand is a good way to end up with a selection which is not shown clearly with the marching ants. This must really screw up PSE users. Sad really, as that product is aimed at less experienced users who might get confused ...

[YrbkMgr]

I guess the Magic Wand is a good way to end up with a selection which
is not shown clearly with the marching ants.




True, but if you've ever done an edge mask, you can't tell for certain there either - or even with this claw method. I almost always create an alpha channel of my selections just to check, out of habit. I like it better than quickmask for inspecting a selection because it's easier to see any feathering.

Frankly, unless I'm using an extremely straight forward way of selecting, I create an alpha channel and inspect it.

<shrug>