Photoshop: Magic Wand tool refinements

  • 1
  • Idea
  • Updated 6 years ago
  • (Edited)
One of my common editing tasks is to increase shadow detail in dark portions of a scan or digital capture. The best tool for carefully selecting shadow areas is the Magic Wand, despite newer selection tools in CS5 and later.

The wand has problems, though. The tolerance setting is opaque, with no control other than specifying a number that doesn't accurately reflect a process that's explained to the user about how the setting is used. When I try to select shadow areas, often far more colors and tones are selected even with a low tolerance than I'd like to have happen. It's clear on the screen that there is a large color difference but for some reason the wand tool won't allow me to discriminate between the differences.

I'd like to see settings that discriminate based on the number of levels that a wand click sample differs from other image pixels rather than an ill-defined "tolerance" number. Also, I'd like more control about how a selection spreads to adjacent pixels, such as a sampling radius. When an image contains noise, a brighter or darker noise pixel seems to accelerate the sampling process and greatly expand a selection beyond what makes sense to the eye. My solution to these problems is time-consuming, and that is to make a copy of the image, apply noise reduction and/or whacked-out curves adjustments to greatly accentuate color and tone differences. Then I make a selection on the processed copy and transfer it back to the original image file. It's a real pain to do this. (Yes, I know how the "sample all layers" and "contiguous" options work. They do not provide enough control for dark shadow editing.)

I also suspect the wand is buggy when editing 16-bit images. It behaves differently and often better when I convert an image to 8 bits. Perhaps this is again related to the way the tolerance setting is used internally.

I hope this gives some insight into how a tool actually works in the real world compared to the conceptual design for it.
Photo of Chris Carvalho

Chris Carvalho

  • 21 Posts
  • 1 Reply Like

Posted 6 years ago

  • 1
Photo of Chris Cox

Chris Cox

  • 20280 Posts
  • 813 Reply Likes
>> I'd like to see settings that discriminate based on the number of levels that a wand click sample differs from other image pixels rather than an ill-defined "tolerance" number.

That is what the tolerance means.

And the magic wand already uses an area to sample the color -- matching the eyedropper area.