Sharpen an image

[Tafflad]

I have read a number of the tutorilas here (and well done guys)

Just been reading up again on Sharpening.

It details using the Unsharp Mask, I have not used that much, having followed the easier route of using High pass Filter
(dup layer - apply High Pass filter - increase until you can just see edges - change to overlay)

Is this still the best approach ... read one comment that says much can be achieved using the Photoshop 'Smart Sharpen' filter instead.


Not arguing with any points, just trying to find what is the current considered approach.

 

[nivla80]

Hi there,

Having encountered ‘problems’ with sharpening itself, I’ve done a fair bit of research and experiments on my images. First off, I found out that sharpening is a beast in itself. There are many different methods to be used for the different mediums in which your images are to be displayed. For now, I’m doing prints and web.

I found out that Smart Sharpen or Unsharp mask works for most part, but the best method (at least what yields the best results for me) is the use of the mentioned techniques, BUT in a systematic way. I shoot a 5D II, and my files are huge. Viewed at 100 percent, after sharpening, they look fine. But as soon as I print it out on a smaller medium, say a 8 by 12, or 800 by 600 px for web, I lose that sharpness. What I’ll do is to resize is a bit at a time, with sharpening applied after every resize. The aim is to get the file to the final resolution/dimension that it will be printed out/uploaded. I also find that having a ‘sharpness halo’ of about 2 pixels on my image on screen yields the best printed results.

Next, I do selective sharpening A LOT. I realized that in my images, certain aspects of it require different amount of sharpening, for example, a beach scene, my foreground rocks and waves requires more sharpening than the clouds. Layer masks help. I’d do 2 passes of USM on the rocks, but only one on the sky/sea, etc. Same thing for underwater images where I’d sharpen the corals more and leave the ocean blue untouched. This helps to get rid of the unwanted noise that MAY be introduced during the sharpening process.

Do visit the links on my signature to see if that’s what you like and if so, that’s the technique that I use above!

HTH!
Alvin

 

[Mark Derail]

I've had great success with this method, don't know if it's documented or not. Based on the fact that a blurred image appears sharp/in-focus when viewed on something smaller.

Similar to looking at a 4x6 print versus an 11x17, the larger size you see what's out-of-focus.

So I duplicate the original image, and shrink it until it is to what I want to publish - print or screen - or half that, if it's already blurred at the desired destination resolution.

Then that "shrunken" image, I use PS to enlarge it to 2x - 8x the original size. On the 2x one, I do any quick corrections necessary, easier with the image being so big. Like details on the face that you want, eyes, eye-lashes, teeth.

So, make it smaller, then make it way bigger, then fix up the bigger, then reduce back to original size.

Use the "modified" version as a layer to the "original" version, and then use the eraser tool to "keep" or "lose" the modified image, that is now a layer on top of the original.

IOW, some parts will be better with this trick, some will look worse. Keep the best of both images.

Remember the destination...seeing individual eye-lashes on a face isn't important on small prints like 4x6 or on a 72dpi screen if the entire body is visible.

IOW, why worry about a fish fin being blurred, if you're not making a poster.

 

[Mark Derail]

I also use the above method for removing things from an image. Say you UW pic of a fish, you have a photobomb of another diver giving you an OK signal.
But you really want a pic of the fish & the corals, minus the diver...

Using the various tools in PS on the original might seem the way to go, but you'll be creating lots of artifacts. Instead, make a copy, work on a TINY version, remove the diver, then blow up the copy back to the original size, and use the new image as a layer on top of the original image.

Then delete from the "on top" layer which is the tiny-diver removed-blown-up everything except where the diver is.

Basically your "removed" diver image has a "better" ocean, pixel-wise, than using a "copy" tool to "paste" bits of ocean here & there on top of the diver.

Plus, this technique is WAY faster, and there are many variations.

Let's say the original image is 4000x3000 pixels, and the diver is 3x as big as the fish, I'll reduce the copied image to where the fish would be in the 10-15 pixel range in size.
Meaning, only about 50-60 pixels of diver to remove, instead of some 1500-2000 pixels.

You're creating artifacts with the stamp tool, in the tiny version, blur the "stamped" area 1x or 2x, then blow up the 400x300 image back to 4000x3000. No more artifacts.

I can remove a diver in about a minute, for simple cases.


Why I like Lightroom for post-processing - I can layer "changes" of various brush sizes, on top of each other, instead of layering images and doing addition or subtractions. I can't remove a diver using LR, but I can easily darken him to oblivion and make the fish stand out.
Sometimes that's good enough.

In LR, have not had good success (like none) with sharpening effects.

Best-bet - always shoot multiple images while on location of the "same" subject. So you can layer up multiple, different, images and keep / lose from the bunch until you have an overall good pic.

So in my example, we have "Bob the Diver" photobombing your awesome pic of a king-sized Queen Angel fish (pun intended).
Simply stay put, wait for "Bob the Diver" to float away, then take another pic of the same spot, at the same angle, as close as possible.

Just so you have "ocean" or "coral" background to add to your original image to paste on top of "Bob the Diver", by layering the two images together.

Or, make a pic of "Bob the Diver" with a giant sea-turtle w/o the giant sea-turtle, and tell him you never saw a turtle...

 

[nivla80]

Sweet! I like the idea of down sizing and removing unwanted elements of the photo and blowing it up again.. Just wondering if I shrink and enlarge the picture, how much quality loss is there, as PS will probably fill in the image with 'extra' pixels...

 

[Mark Derail]

PS does a very good job. It's a great way to "lose" detail in some places, and "keep" detailed from the original.

I saw one tutorial where such a technique is used to make "nice" skin w/o making the person look like a barbie doll skin.

The technique works wonders on straight lines, like if you want to remove a wire, a pole, your tripod...

Sweet! I like the idea of down sizing and removing unwanted elements of the photo and blowing it up again.. Just wondering if I shrink and enlarge the picture, how much quality loss is there, as PS will probably fill in the image with 'extra' pixels...