Sulman From United States of America, joined Mar 2004, 2041 posts, RR: 28 Posted (11 years 6 months 2 weeks 2 days 12 hours ago) and read 3315 times:
I've been exploring photoshop some more and I've got some questions about sharpening technique. I typically use 30/0.3/0 for percentage amount/pixel radius/threshold and it's served me well (thanks Justin!), however I'm still not sure it's ideal in some cases.
My interest was prompted when I read this informative article here:
I found it helpful as, aside from having a very basic knowledge of what USM does, I didn't have a clue what the settings really meant. Having said that, I looked at my own settings, and dug up an old thread here, and found most people prefer very low figures, but in particular a threshold value of '0'. Essentially, this means that Photoshop will apply the sharpening algorithm to all pixels where a difference is detected after creating the mask; no matter how small that difference is.
That was enlightening for me, as I had noticed on some of my images a very fine - but noticeable - saw-tooth effect on low contrast edges, such as white fuselages against the sky, or wing leading-edges against any background. I'm not talking about the USM stegosaurus you get after oversharpening, but something more subtle. Increasing the threshold did seem to help with this quite a bit.
Any techniques people want to share? Do you just avoid USM 'problem' areas and process parts of the image individually? People do get some amazing results and I thought it would be useful to exchange some tips.
It takes a big man to admit they are wrong, and I am not a big man.
Ckw From UK - England, joined Aug 2010, 813 posts, RR: 14
Reply 1, posted (11 years 6 months 2 weeks 2 days 12 hours ago) and read 3286 times:
This is such a difficult area ... to get good results is pretty straight forward, but to get optimum results, well that's another matter! Moreover, as the best results are also dependent on camera type (and in-camera settings), file processing method (in the case of RAW) and required output size, the ideal formula is going to be pretty specific to any individual photographer and picture.
For this reason, 3rd party tools such as FocalBlade are very useful since they a) analyse the actual image being sharpened to determine an optimum algorthm and b) provide simple slider controls to fine tune things like haloing (which would require very time consuming masking and channel work in PS).
I have been through the process of writing reasonably complex actions for optimum sharpening, but, apart from the intellectual interest, this is really a waste of time. I don't worry about it too much anymore.
For everyday work, I'll use a standard USM formula - for my 1Dii, its something like 300-400/0.3/0 for A.net size images; 300-400/0.6/3 for A4 size prints (note that unlike the 10D/D60/300D which always apply some sharpening, even with sharpening set to 0, the 1Dii applies none).
For very critical work, I'll use Focalblade.
Subject matter makes a difference as well - in general, with aviation pics, we just want the image "sharp". However, with portraits, such an approach can be unflattering - we want sharp eyes, but smooth skin, so here selective sharpening becomes necessary, though once again, products like Focalblade make this much easier.
ChrisH From Sweden, joined Jul 2004, 1136 posts, RR: 14
Reply 2, posted (11 years 6 months 2 weeks 2 days 12 hours ago) and read 3276 times:
On a 1024x image I usually start with 80%/0.3/3, and then adjust percentage to what looks good. I do this in LAB-mode in lightness channel. If jaggies appear, Either sharpen the image in a new layer, then use erase tool to "paint away" the jaggies, using low flow (press 3 on keypad for 30% flow, 4 for 40% etc.). Or you can use the historybrush to accomplish the same thing really. Also I use quickmask to select the aircraft or details I want sharpened, leaving skies and background untouched.