Wednesday Q&A - Making Better Selections With Adobe Photoshop

I've seen a lot of frustration with friends and readers trying to make selections using Adobe Photoshop CS5's Content Aware Fill.  Today's image shows three versions of the same image.  It's not on the blog because it's some wonderful image.  It's here as a "teaching tool".  It's the same shot, duplicated twice and spread out across a canvas.  The first (left) shows the "as shot" image.  The middle version show a typical Content Aware Fill  result, trying to take out the middle "R" and the number "7".  The third (right) shows a better result using a modified version of Content Aware Fill.  Why someone would want to remove the "R" and the "7"?  I have no idea, I'm just trying to illustrate a point.  I didn't come up with the secret method of making a better fill.  I saw it in a video from Adobe.  I've been using it, tweaking the way I work with it, and have developed a strong confidence in being able to extract exactly what I want.  To find out what the trick is, hit the "Read More".

The big deal is using a Layer Mask.  Make a copy (Ctrl J) of the background image.  Add a Layer Mask (the front loading washing machine looking icon at the bottom of the Layers Panel).  Mask off whatever you don't want Content Aware Fill (CAF) to look at.  Make any sort of selection you are comfortable with.  i.e. The Loop Tool (L) or the Marquee Tool (M).  This way you're preventing CS5 from "looking" at areas that might produce false fills.  Look at the center image.  Part of the "R" was picked up, as was a double part of the "0". 

After making a selection of the "C",  the right "R" and the "10" and "2" you would add a Layer Mask.  This will create a Mask that "hides" those parts of the image.  Go back to the image itself (not the Layer Mask) and select the middle "R" and the "7".  Now do a Content Aware Fill on that selection.  Because CS5 can't "see" the other letters and numbers it'll take all its information for doing the fill from areas that are still "visible" to CS5. 

In the case of today's image, the bright area at the top doesn't come into play.  Therefore there's nothing that has to be done to that area.  If it had it would be a simple matter of adding that area to the Mask. 

The "morale" of the story to making a tricky Content Aware Fill is to "show" CS5 only what you want it to use to that matches up well with what you're trying to fill.  Leaves with leaves, fences with fences, etc.

Try it, it'll help.