Posted by Jody Grigg on October 4, 2011 at 12:05 PM
Like they say a silly question is a question not asked...so here goes...
How do you Smart Sharpen? What I really want to know is do you do it based on the Smart Sharpen box at 100% or how it looks on the main image. I know if I have an image sized at say size 'X' when I go to the navigator window put it at 29% it is the exact size that the print would be. Then I open Smart Sharpen tool make my minor adjustments and look how the main image changes until I like the look. Recently I read that you should look at the 100% box attached to the tool, I tried this way but my main image always looks a lot softer and never sharp. When I do it my method the 100% box does not necessary look that good but I have never yet had a problem with a print made on either paper or canvas based on this and I have printed upto 40"x60" on my method. Am I doing this all wrong, or is there method to my madeness. I am wanting to learn so let me have at it. Thanks in advance for any input.
After smart sharpen is open I use lens blur. I was taught see how the image looks using the preview button at various amount and radius settings. I also use the more accurate setting. Radius and amount are reciprocal to each other, ie high amount low radius and vice versa. Make sure you look at the image at 100% and 50% to help determine which setting work best to your eye, again utilizing the preview button. Use the least amount of sharpening you need to to get the job done. Too much introduces artifacts
If you learn how to control the "Blend If" sliders in PS and sharpen in either LAB L mode or RGB Luminosity, it pretty much makes Smart Sharpen irrelevant (or incomplete) IMO. I'm producing a DVD series on this soon. But I would ditch Smart sharpen. A simple formula would be to duplicate a layer in PS and change it to Luminosity mode, and go to the
Layer Style "Blend If" sliders and create your highlight or shadow fall off there on the fly (like 0/30 on black and 200/255 on white maybe). For high detail shots like landscape, sharpen with USM (still the best sharpening tool in the known universe) with amount at 500, threshold at 0, and then inch the radius up very slowly. If the image is 300ppi or above view the image at 25%. If the image is 200PPI or below view at 50%. Never at 100%. Go slightly, slightly crispy on screen since prints come out softer than monitors. Lastly look into grain simulation for added texture (but not to make the image look grainy) as a last step. It camouflages digital artifacts making a much more organic look to the print, when done right.
Lastly, LOTS of test prints!
Have fun.
If you're happy with your results, especially at all sizes, then listen to everybody for intellectual reasons but then keep doing what you're doing!! You're gonna get a zillion different "answers" but they'll likely be more "opinions" because while there is clear science behind sharpening for sure, how one uses it is a matter of personal choice rather than a single "correct" way.
For example, Tim Grey states that Smart Sharpen in the new CS5 is the bees knees and the incorporated algorhythms have virtually eliminated the need for Unsharp Mask, all the arcane LAB mode, luminosity, etc etc as long as you choose the Lens Blur option and not Gaussian Blur.
For what it's worth, I used to use unsharp mask exclusively, sometimes doing the luminosity mode dance, until I watched his DVD on sharpening and then switched to Smart Sharpen with great results. Life is much simpler now and even though on a purely scientific level there may be better ways, if the results are not significant enough for me to see in my prints, I don't see the need to pursue it. I simply set my image on screen to "print size" (or close to it) before initiating SS. The window box in SS is at 100% as noted so I see no need to also have the image at 100%. By having it close to print size, I judge the result before and after by using both the image with preview button and left clicking in the 100% box. Almost always I make my final determination by how the image at print size looks and only use the 100% box to be sure that it has not gotten too noisy, have objectionable halos or other issues. I also always go just a teeny bit too much on screen to make up for the slight ink bloom softening of inkjet printing.
Posted by David Bostedo on October 5, 2011 at 8:57 AM
You didn't mention your Photoshop version... if it's CS3 or earlier, viewing at intermediate percentages will definitely look softer. I.e. if it's not 25, 50, or 100 percent, you aren't seeing things as well as you should. They improved the scaling at the other percentages for CS4 and CS5. So your 29% is particularly a bad way to view sharpening if you have CS3 or older.
That being said, one of the nice things Smart Sharpen does for you instead of USM is pin the change in contrast around edges so that it won't really go brighter or darker than neighboring pixels. This largely prevents halos as you turn the sharpening up. But I've used smart sharpen, LAB, USM, high-pass, various RAW conversion sharpening, and a couple other programs (but never an aftermarket plug-in) and I've gotten good results with all of them. So I agree with John Reed about finding something you like and that gives good results and just going with it.
"So I agree with John Reed about finding something you like and that gives good results and just going with it."
Same here, Jody. If you have a procedure for sharpening for print and another for sharpening for the Web that give you the best results--stick with it.
I set the PS ruler to 'inches' and adjust the view so that one inch on the screen equals one inch on a physical rule (or very close to it). I use CS5 USM on a copy of the background layer. The blending mode is set to Luminosity. Amount, Radius and Threshold are image dependent, and I apply USM in stages and fine tune as I go along. For printing, I find that if the image looks a little 'crunchy' on screen, the print is sharp. I often 'View Actual Pixels' (100%) to check for halos and noise. Next, I make a proof print to verify that all is well.
Having said all this, I may try Smart Sharpen using Lens Blur just for curiosity's sake. Research continues...
--P
Posted by John Mumaw on October 5, 2011 at 1:10 PM
Whichever method you choose, be sure to apply on a separate layer in Luminosity blend mode (as has been noted here).
This way you can use layer masking (rarely does the entire image need sharpening) and opacity adjustments for maximum control.
Discussion
6 repliesHow do you Smart Sharpen? What I really want to know is do you do it based on the Smart Sharpen box at 100% or how it looks on the main image. I know if I have an image sized at say size 'X' when I go to the navigator window put it at 29% it is the exact size that the print would be. Then I open Smart Sharpen tool make my minor adjustments and look how the main image changes until I like the look. Recently I read that you should look at the 100% box attached to the tool, I tried this way but my main image always looks a lot softer and never sharp. When I do it my method the 100% box does not necessary look that good but I have never yet had a problem with a print made on either paper or canvas based on this and I have printed upto 40"x60" on my method. Am I doing this all wrong, or is there method to my madeness. I am wanting to learn so let me have at it. Thanks in advance for any input.
Layer Style "Blend If" sliders and create your highlight or shadow fall off there on the fly (like 0/30 on black and 200/255 on white maybe). For high detail shots like landscape, sharpen with USM (still the best sharpening tool in the known universe) with amount at 500, threshold at 0, and then inch the radius up very slowly. If the image is 300ppi or above view the image at 25%. If the image is 200PPI or below view at 50%. Never at 100%. Go slightly, slightly crispy on screen since prints come out softer than monitors. Lastly look into grain simulation for added texture (but not to make the image look grainy) as a last step. It camouflages digital artifacts making a much more organic look to the print, when done right.
Lastly, LOTS of test prints!
Have fun.
For example, Tim Grey states that Smart Sharpen in the new CS5 is the bees knees and the incorporated algorhythms have virtually eliminated the need for Unsharp Mask, all the arcane LAB mode, luminosity, etc etc as long as you choose the Lens Blur option and not Gaussian Blur.
For what it's worth, I used to use unsharp mask exclusively, sometimes doing the luminosity mode dance, until I watched his DVD on sharpening and then switched to Smart Sharpen with great results. Life is much simpler now and even though on a purely scientific level there may be better ways, if the results are not significant enough for me to see in my prints, I don't see the need to pursue it. I simply set my image on screen to "print size" (or close to it) before initiating SS. The window box in SS is at 100% as noted so I see no need to also have the image at 100%. By having it close to print size, I judge the result before and after by using both the image with preview button and left clicking in the 100% box. Almost always I make my final determination by how the image at print size looks and only use the 100% box to be sure that it has not gotten too noisy, have objectionable halos or other issues. I also always go just a teeny bit too much on screen to make up for the slight ink bloom softening of inkjet printing.
That being said, one of the nice things Smart Sharpen does for you instead of USM is pin the change in contrast around edges so that it won't really go brighter or darker than neighboring pixels. This largely prevents halos as you turn the sharpening up. But I've used smart sharpen, LAB, USM, high-pass, various RAW conversion sharpening, and a couple other programs (but never an aftermarket plug-in) and I've gotten good results with all of them. So I agree with John Reed about finding something you like and that gives good results and just going with it.
Same here, Jody. If you have a procedure for sharpening for print and another for sharpening for the Web that give you the best results--stick with it.
I set the PS ruler to 'inches' and adjust the view so that one inch on the screen equals one inch on a physical rule (or very close to it). I use CS5 USM on a copy of the background layer. The blending mode is set to Luminosity. Amount, Radius and Threshold are image dependent, and I apply USM in stages and fine tune as I go along. For printing, I find that if the image looks a little 'crunchy' on screen, the print is sharp. I often 'View Actual Pixels' (100%) to check for halos and noise. Next, I make a proof print to verify that all is well.
Having said all this, I may try Smart Sharpen using Lens Blur just for curiosity's sake. Research continues...
--P
This way you can use layer masking (rarely does the entire image need sharpening) and opacity adjustments for maximum control.