Help with sharpening photos

[KirkT]

I think I have most of my digital workflow down pretty well, except for sharpening (and nice neutral colors, but that's another thread).  Both my digital and scanned photos could stand to be a little snappier.

What are your techniques for getting a nice sharp photo?  Do you use USM?  Is it standard across all types of photos or is one better for portraits vs street vs landscape, etc.

Thanks.

Kirk

[Mike Kovacs]

Photokit Sharpener Pro plugin

[jake]

Nik is the plugin that a lot of the pros use. I don't have a lot of plug ins for Photoshop. Actually, I don't have any. I just use Photoshop. CS3 added a much improved advanced menu to their sharpening tool, giving a bunch of different ways that sharpening can be applied.

However, I generally do this:

No sharpening at input. That means turning off sharpening during the scan or RAW step. Most of the TWAIN-based programs are pretty clumsy with their sharpening. Photoshop is much better.

Apply sharpening after all the other adjustments. Once I get color, exposure,  and such set up, then I work on sharpening. If I was smart and did the right thing with the camera, I have found that 100%/2/1 are the settings that I use the most on the unsharp mask dialogue. I always forget what the other two things are - radius & pixels or some such.

Lightroom seems even better at sharpening gently than Photoshop, so if I have a batch of photos that need just a touch, I use Lightroom and paste the sharpening settings to all the photos. Quick and seems to do a good job.

What is sharp on screen may not be when you print, so I print a proof then adjust sharpening accordingly.

Color cast: You have to use curves. I start with the blue channel first, usually subtracting it from the shadows and adding or equalizing it in the highlights. Then I bring up the greens in the mid-tones (grass is typically a mid-tone) and raise red a bit in the shadows while keeping it down in the sky or highs. That typically gives me what I remember as a Kodachrome-look, which is what I think of when I think of reality. On bad shots, then sometimes I have to go to "Selective color" and adjust the neutrals/whites/blacks to get everything to behave. But really at that point, I should just park that negative as improperly exposed.

[Alan Gage]

I used to agonize over sharpening and finally came to the conclusion that I could spend 10 minutes fiddling with it or just apply the same level to every photo and it would turn out pretty darn good.

For most of my images, which are nearly all just web based, I apply something like 75/1/1 of USM in Photoshop.

If I'm making a print then it usually takes more sharpening. I've found the larger the print the more sharpening it can stand/needs. I generally only increase the amount of sharpening and leave the other two numbers at "1".

So far this has worked well for me. I'm sure there are better ways and I'm sure that some of the pics could look better; but I've decided for me that this way is fine.

Oh, and like Jorn said. Sharpening is the last step, the previous step being resizing for final output.

Alan

[Madrigal]

Yikes!! Boy oh boy were my USM settings all wrong. Jake- for whatever I was scanning just now in PSE- 100/2/1 did the trick.

Thanks!!! Now all I have to do is go re-scan everything :2yippee::2yippee:

[Madrigal]

My old USM setting.

The new USM setting.

:thumbsup2:

[jake]

Eggsalent.

[edthened]

Och Picasa 3, press sharpen wunce, dun,  an it costs zilch, so ther :muching_out:

[martolod]

Och Picasa 3, press sharpen wunce, dun,  an it costs zilch, so ther :muching_out:

yep me too.

[Graham Serretta]

Photokit Sharpener Pro plugin

What Mike said.

[bruce]

I use Gimpshop, and find that it helps if I creep up on the sharpening. I scan at full resolution with no sharpening. I then make any levels/curve adjustments. I then adjust the sharpening to what looks good; I cut these settings by a third, and sharpen. I then resize the image to the size I want, and after doing that, sharpen again with the same settings. If I am going to save the image as a jpeg for the web or such, I will do so, and then sharpen a third time with the same settings.
 This seems to give me better results than just doing it all at once. If I do it all at once I tend to over sharpen the images, a look I don't really like.

[Tom Hildreth]

For my negs and slides I scan without any sharpening/USM and save as TIFF. When I have selected an image I want to print, or post, or use in some way, I begin my editing process. Toward the end of this process, when I have the image in the final resolution I want, I then normally apply USM at around 40%/0.5 pixels or 1.0 pixels/threshold around level 8.
 
 
Do check out the high pass technique here. I am very glad I found this technique, and turn to it whenever I'm not happy with my USM results.

[Harry Haller]

very interesting thread.

I learned the highpass technique here, but preferred converting lo Lab and shapening the L channel - if you did over sharpen, you'll see white somewhere

I cannot remember exactly why I don't use highpass anymore.

Perhaps because I started exploring NeatImage, and also Nik specially to reduce noise - which is somehow contradictory as they often judge grain as being noise. OTOH, I gave up using the USM options in CS3 because they do increase noise (nonetheless, I still find LCE settings in CS3 very useful).

I'm currently exploring Fred Miranda Intellisharpen II, and resizing with SI Pro 2 - and finding them quite usefuful.

regarding the scanner:

I have a second hand Nikon 4000ED, with ICE3. its software interface (NikonScan 4.0.2) has curves/levels/usm. I regularly use curves/levels at each channel, and last at the global RGB curve which improves contrast with minimal adjustment. I do use a little of usm at Nikon 4.0.2 - threshold and radius at default, intensity 'round 10 - 15%. these settings always fall short. I've been advised on the contrary: not even touching a single parameter in NikonScan, and leave everything to CS3. must confess that either what I am doing and the advice on the contrary, don't fit into a complete logic realm - both have loose ends.

last, I regularly commit an error: I do sharpen before I know what I will be doing with that particular image. after I get the color balance, histogram, etc. right (nearly right for that particular state of consciousness, in fact) - I try to match the perceived sharpness of my Provia 100F / Velvia 50. things are coming out better it seems: when I load my 16bit .tiff done in gamma 2.2 and europe prepress 2 at D65 using a graphic series CRT monitor, into a Mac LCD monitor previously set at 2.2 and D65 - I see differences as expected. I've been lucky when printing large, too - but not always when printing through a Frontier.

hope my English can be understood, I'd like to read further comments. and I feel I must thank five people who have helped or confused me this far.

[martolod]

hope my English can be understood,

nothing wrong with your english....it's better than my french

the luminous landscape tutorials have been  a favorite of mine for quite some time.most of the techniques i use have come from there in particularly the use of layers.
once you have mastered the art of layers you well on the way of using photoshop more effectively.

[Harry Haller]

nothing wrong with your english....it's better than my french

the luminous landscape tutorials have been  a favorite of mine for quite some time.

lol  :cool:

I've found these very helpful too. thanks Chuck for this link, some five years ago iirc.