Serious selection issues in CS2/CS3!

CC
Posted By
Captain_Cook
Jun 16, 2007
Views
363
Replies
11
Status
Closed
I’d like to see if anyone can explain this strange phenomenon to me…

TEST IMAGE… <http://tinypic.com/4tr6d7c>

Try this if you would…

Starting with the my test image, create a selection inside the red square with the rectangular marquee tool, then select the magic wand tool and set it to Tolerance 100, Anti-alias, Contiguous.

Now right click on your selection and click on "Similar".

The result should be a nice full selection of the red square… <http://tinypic.com/4qh0y2q.jpg>

Here’s where things get weird.

Starting over, click on the center of the blue square with the magic wand tool (same settings as before). Now holding the shift key, again make a selection inside the red square with the rectangular marquee tool.

Finally reselect the magic wand tool and click similar from inside your selection. The result is different from what you’d expect… <http://tinypic.com/4v6362w.jpg>

Note the red square has not been fully selected despite instructing Photoshop to select pixels similar to the existing selection.

Needless to say I’m getting frustrated by this issue since I depend on Photoshop for a living.

Thanks

Must-have mockup pack for every graphic designer 🔥🔥🔥

Easy-to-use drag-n-drop Photoshop scene creator with more than 2800 items.

DP
Daryl_Pritchard
Jun 16, 2007
Well, I have no explanation for this, but I’m guessing that expanding the selection based upon similarity apparently must be taking into consieration the color of all selected pixels as perhaps some sort of average value. Or maybe it is simply that Similar doesn’t work this way for adding to a selection. Perhaps it wouldn’t work for what you’re doing, but rather than use the marquee tool at all, why not just use the magic wand to select all of the blue initially, then expand to include all of the red via a Shift+Click of the magic wand in the red area?

Regards,

Daryl
DM
Don_McCahill
Jun 16, 2007
I can’t understand the question. Are you expecting select similar to select colors that are similar to both the blue and the red? I’m pretty sure it doesn’t work that way. It will only select colors that are similar to one color, not two. Adjusting the tolerance level on the tool might give you the correct result occasionally, but not reliably.
DM
Don_McCahill
Jun 16, 2007
I can’t understand the question. Are you expecting select similar to select colors that are similar to both the blue and the red? I’m pretty sure it doesn’t work that way. It will only select colors that are similar to one color, not two. Adjusting the tolerance level on the tool might give you the correct result occasionally, but not reliably.
CC
Captain_Cook
Jun 17, 2007
Don McCahill – 5:29am Jun 16, 07 PST (#2 of 2)

I can’t understand the question. Are you expecting select similar to select colors that are similar to both the blue and the red?

Yes I am.

Bump!
GA
George_Austin
Jun 17, 2007
Captain Cook,

I am assuming your red swatch is 255/0/0. Just change its blue component to any non-zero value (255/0/1 will do). Now your prescribed action will result in selection of the entire "red" swatch.

George
EH
Ed_Hannigan
Jun 17, 2007
Cap’n,

I’m trying to select colors similar to the existing selection (red OR blue).

But the existing selection is NOT red or blue, it is red AND blue.
GA
George_Austin
Jun 17, 2007
Well, this is getting interesting. My previous assertion was for a specific geometry and not a generalization and I retract it.

What I now find is that the whole enchilada gets selected if the selected area within the red swatch is greater than 11% of the swatch area—that is, greater than one-ninth the area—meaning the side of an interior square selection is greater than one-third the side of the patch if the patch is square. If the interior selection’s area is less than one-ninth of the patch area, the selection is not extended. Further, the (shall we call it "threshold") area is independent of the wand’s tolerance setting. If this holds up, the next obvious question is why such a seemingly arbitrary constraint? The algorism for "Similar", please!!
S
Silkrooster
Jun 17, 2007
All I know, is that I would take the least common denominator and select just the black and invert the selection. Why it does for you what it’s doing, I don’t have a clue.
CC
Captain_Cook
Jun 17, 2007
George Austin – 11:38pm Jun 16, 07 PST (#6 of 7) Edited: 16-Jun-2007 at 11:40pm

Well, this is getting interesting. My previous assertion was for a specific geometry and not a generalization and I retract it.

What I now find is that the whole enchilada gets selected if the selected area within the red swatch is greater than 11% of the swatch area—that is, greater than one-ninth the area—meaning the side of an interior square selection is greater than one-third the side of the patch if the patch is square. If the interior selection’s area is less than one-ninth of the patch area, the selection is not extended. Further, the (shall we call it "threshold") area is independent of the wand’s tolerance setting. If this holds up, the next obvious question is why such a seemingly arbitrary constraint? The algorism for "Similar", please!!

Thanks for the detailed reply George. I’m trying to understand how the tool works since the documentation is very limited.

I know there are a million other ways to accomplish selections for but my line of work the magic wand is the fastest and most convenient. I just wish it would perform how I expect it to.
CC
Captain_Cook
Jun 18, 2007
BUMP!
CC
Captain_Cook
Jul 16, 2007
Bump

MacBook Pro 16” Mockups 🔥

– in 4 materials (clay versions included)

– 12 scenes

– 48 MacBook Pro 16″ mockups

– 6000 x 4500 px

Related Discussion Topics

Nice and short text about related topics in discussion sections