DSLRUsers.com

Ok, dumb question time.

Why do all my images look better with unsharp masking? Is there a setting I haven't found in the camera to do this? What am I missing?

Thanks,
Brett/skippy

In short, nup there is no camera settings you've missed.

Unsharp mask consists of making a separate luminosity layer, applying a gaussian blur to that layer, and then overlaying it with the original and highlighting the edges to make them more pronounced. All this is done for you by photoshop. The complexities of which are why simpler sharpening algorithms do not work as effectively.

If you've ever been unfortunate enough to shoot a Canon, you will note just how lacking in finer details your images contained prior to applying USM - it is then that all of a sudden details seem to 'pop' from nowhere.

Because of the superior design compromise employed by the talented engineers at Nikon, you end up with a sharper image out of camera from your D70, but still USM is beneficial in extracting edge contrast.

Basically, it's the way our eyes function. Humans perceive depth by virtue of having two eyes slightly apart on our head. The camera removes this cue for depth perception. Thus, the eyes are now reliant upon only edge contrast to differential where one object ends and the other begins, to separate foreground from background. USM assists here quite nicely.

Also helps is depth of field. Having infinitely DOF in an image generally confuses the viewer and overwhelms the eyes, for we do not normally see everything in our field of vision rendered sharply - only those which the eyes focus on. However our eyes have the advantage of scanning portions of a scene 40 times a second or more, and to conclude that only the salient object is rendered "in focus" to us. Hence the limited depth of field employed for wider lens apertures in some way mimicks our visual system by artifically restricting the scanning process to only the subject which we want the eye to focus on.

Much of this has nothing to do with your original question, which is probably a sign of my bedtime. Nighty night.

Onyx wrote:In short, nup there is no camera settings you've missed.

That's nice, but bugger. Was hoping there was an easy way to fix it without ... how do I put this ... dicking about with PS every time. Yes, that's it.

So the Gaussian blurring on the second layer is affected by the radius of the masking? Usually I just go with the defaults, which seem to be amount ~93%, radius 2.0 pixels and threshold 0 levels.

The radius setting seems to work almost like contrast, and I can't really describe what the threshold does except make it less effective in a strange way. Can you explain the details behind these settings?

Oh, and nighty night...

My favourite settings for USM are:
Amount: between 100% and 300%
Radius: between 0.3 and 0.5 pixels
Threshold: 0

Rinse and repeat if necessary.

This gives a nice crispness without being overly obvious.

Cheers
Dean

Unsharp mask is a leftover from the old days of film and that is why it has such a strange name.

This site has a nice write up on USM

http://www.luminous-landscape.com/tutorials/understanding-series/understanding-usm.shtml

Chris

It is interesting that there are several favourite unsharp mask settings.

Deano, I don't quite understand how you can have between 0.3 and 0.5 pixels as your radius. Wouldn't it have to be a number of pixels rather than a part of a pixel?

The threshhold of ) means that every pixel will be effected. I have seen a suggested threshhold of 3 or 4 to avoid oversharpening artifacts.

I have also seen an amount of 15% which seems very low - your 100% to 300%, at the lower end anyway, seems more the go.

But I don't know - I am just interested in the widely ranging suggestions

This whole sharpening issue is very important given that the low pass filter in front of the CCD softens the image, and I am determined to get to the bottom of it...

Greg B wrote:Deano, I don't quite understand how you can have between 0.3 and 0.5 pixels as your radius. Wouldn't it have to be a number of pixels rather than a part of a pixel?

I too do not understand how a radius of less than one pixel works but it does.

I got this setting info from Bill Bates at http://www.pbase.com/slowpokebill

Cheers
Dean

Those unsharp settings are fairly well explained here:

http://www.bythom.com/sharpening.htm

I use the same values for all but amount (where I use a value as low as 60) as a starting point.

Cheers,

Matt

The use of smaller radius (<1 pixel) are merely to avoid the halo effect (happens when oversharpenned).
I think there are enough links to external sites here to keep anyone interested entertained for hours.

Deano wrote:I too do not understand how a radius of less than one pixel works but it does.

Well colour me surprised!

Further investigation is clearly on the cards.

thanks guys