How do you sharpen? Whats your workflow?

[Shootar401]

Mine is pretty simple. 90% of all my work is printed at 11x14 so I'll use that as a sample.

- Scan my negs on my V700 with Epson Scan (SilverFast is TERRIBLE!) at 4800 dpi.
- Open in Photoshop.
- Do my straightening and cropping, apply any printer profiles.
- Resize my image to 300dpi, reduce 50%.
- Apply USM at 50/1.0/0
- Resize the image again to 50%.
- Apply USM at 50/1.0/0
- Do a final resize to 11x14
- Save to TIFF to import into CaptureOne for adjustments. (because we all know PS is useless)

This gives me, what I think is the best sharpening for almost all of my work. Much better than using the high pass filter and blending mode.

Whats yours?

[bob carnie]

PS is useless??

So you sharpen before you make colour adjustments?

[Kirk Gittings]

A bit of trolling here. Silverfast is not terrible. You obviously have not learned to use it. Having said that I've switched to Vuescan

[Shootar401]

PS is useless??

So you sharpen before you make colour adjustments?

Yes, I usually get my color spot on at time of scanning, at the most I'll do some curves and levels in C1

[Preston]

PS is useless??

So you sharpen before you make colour adjustments?

Since I don't have CaptureOne, Lightroom, or...I use PS-CS6 exclusively. I would hardly call it useless.

I am curious as to why one would sharpen before making adjustments? Sharpening is the very last step I take, whether for print or the Web.

I do use the 'Sharpen' filter in Vuescan and then save my scan as a PSD 'master file'. All adjustments are made to the master file: It is never sharpened. If I want, say, an 8x10, I duplicate the master file, save it under a new name at 360 dpi (for my Epson 2400) and then sharpen that image as needed.

--P

[MIke Sherck]

- I use a cable release
- I always use a tripod
- I focus carefully
- I wait a moment for the camera to settle after pulling the dark slide and setting the shutter
- I'm careful not to jiggle the camera when I trip the shutter

;) Mike

[Kuzano]

All very good tips for getting a very sharp capture, which is a leg or two up on anything one can do in the scan, or post processing.

I will add one tip that I learned years ago, and was given some incredible number for the percentage gain in stability... Have learned to trust that tip and claim implicitly.

I always add a hook or eye to the bottom of the center post of my tripod whenever I get a new one.

1) I carry a net style shopping bag in my gear, which I fill with rocks on site and hang from the bottom of the center post. OR

2) Upon embarking out to shoot, I fill a couple of two gallon milk jugs with water and hang one or two from the center post of the tripod.

Naturally, no wind, and let the liquid option settle down before activating the shutter. But, isn't that why we shoot large format anyway. To fill the time void in our lives by not having much else happening anyway?:cool:

[paulr]

My goal in capture sharpening is to get the best pixel-level sharpness and s/n ratio possible, while still keeping the file looking natural at 100%. If it's digital capture I use Lightroom, which provides all the needed tools in one panel. This step can be done by eye at 100%, and will vary from image to image, lens to lens, etc... Typical settings are 0.6 to 0.7 pixel radius, 40 to 60%, "detail" set to around 25, masking set to 0. The detail slider is the most interesting; it controls the ratio between the two sharpening algorithms. At zero it's all unsharp mask; at 100 it's all blind deconvolution. The setting that works best depends on how well the deconvolution algorithm's point spread function matches the blur in your image. If it's a bad match, you get a ton of noise. If it's a good match, it works like magic.

For scans, I do this in photoshop, using a sharpening layer. The image layer is duplicated, with blend mode set to luminosity. The blend sliders are set to exclude, with a lot of feathering, the darkest shadows and brightest highlights. I use the smart sharpen filter, with the blur set to "lens blur," and other settings similar to the ones mentioned above for lightroom. Using the lens blur option adds deconvolution to the sharpening algorithm and usually (but not always) gives smoother looking results than unsharp mask. Smart sharpen used with gausian blur is identical to unsharp mask.

I also do a bit of local contrast enhancing. In lightroom this is done by turning up the clarity slider; it photoshop by running a pass of unsharp mask at large radius (180 to 250 pixels) and a very small intensity (10 to 15%).

Output sharpening is also done on a sharpening layer in PS; the radius is set to equal 1/10 mm at print size, to correspond to 5 lp/mm, unless the print will only be viewed farther than 10". For every additional 10" minimum viewing distance, add another 0.1mm. Amount of sharpening requires some judgment; I find that viewing on screen at 50% scale for 360ppi files let's me get it right most of the time. A file properly sharpened for printing will look significantly oversharpened at 100% on screen.

[Walter Calahan]

I do a very slight global unsharp mask via Silverfast. I do all the rest of my sharpening, using the high pass filter technique applied to only those areas of the image needing sharpening, at the very end after all the adjustments, cropping, and sizing.

[Shootar401]

How would one go about doing this? I tried duplicating the layer, but after setting the blending mode to luminosity I'm lost



For scans, I do this in photoshop, using a sharpening layer. The image layer is duplicated, with blend mode set to luminosity. The blend sliders are set to exclude, with a lot of feathering, the darkest shadows and brightest highlights. I use the smart sharpen filter, with the blur set to "lens blur," and other settings similar to the ones mentioned above for lightroom. Using the lens blur option adds deconvolution to the sharpening algorithm and usually (but not always) gives smoother looking results than unsharp mask. Smart sharpen used with gausian blur is identical to unsharp mask.

[Andrew O'Neill]

- I use a cable release
- I always use a tripod
- I focus carefully
- I wait a moment for the camera to settle after pulling the dark slide and setting the shutter
- I'm careful not to jiggle the camera when I trip the shutter

Mike

... then add an unsharp mask overtop of the negative...:)

[Richard Mahoney]

I do a very slight global unsharp mask via Silverfast. I do all the rest of my sharpening, using the high pass filter technique applied to only those areas of the image needing sharpening, at the very end after all the adjustments, cropping, and sizing.

I tend to leave all digital file prep to the fellow who does my prepress -- he has more experience than I'll ever have. Initial sharpening is via the drum scanner. Secondary sharpening is via Photoshop after all other changes.


Best, Richard

[paulr]

How would one go about doing this? I tried duplicating the layer, but after setting the blending mode to luminosity I'm lost

There's a pulldown menu for blend modes. Selecting "luminosity" means that your sharpening will work only on the luminosity channel and will leave the color channels alone. This keeps you from creating or accentuating any color artifacts.

The blend sliders are not very intuitive ... I'll see if I can find a web tutorial to link to for you. I'd just make it more complicated than necessary by trying to describe. The general idea is to exclude the shadows and highlights. This keeps you from pushing any pixels into clipping, and helps keep you from exaggerating shadow noise.

The biggest benefit to this kind of workflow is that all your sharpening is non-destructive. You're only sharpening this layer, which can be discarded and remade if necessary. You're not touching the original pixels.

If Adobe ever gets it together and invents a sharpening Adjustment Layer, most of this will be taken care of for us.

[bob carnie]

I sharpen selectively on the L channel for this very reason.

There's a pulldown menu for blend modes. Selecting "luminosity" means that your sharpening will work only on the luminosity channel and will leave the color channels alone. This keeps you from creating or accentuating any color artifacts.

The blend sliders are not very intuitive ... I'll see if I can find a web tutorial to link to for you. I'd just make it more complicated than necessary by trying to describe. The general idea is to exclude the shadows and highlights. This keeps you from pushing any pixels into clipping, and helps keep you from exaggerating shadow noise.

The biggest benefit to this kind of workflow is that all your sharpening is non-destructive. You're only sharpening this layer, which can be discarded and remade if necessary. You're not touching the original pixels.

If Adobe ever gets it together and invents a sharpening Adjustment Layer, most of this will be taken care of for us.

[paulr]

I sharpen selectively on the L channel for this very reason.

Same result, but luminosity mode allows you to work in RGB. I realize some people prefer to work in LAB for other reasons.

[Peter Gomena]

This thread is why I love this forum. I've learned more here today than in the past year from other sources. Thanks to all for the enlightening discussion.

[bob carnie]

Quick question, when you work with the luminosity blending mode are you able to see the image in BW ?

Same result, but luminosity mode allows you to work in RGB. I realize some people prefer to work in LAB for other reasons.

[Ari]

http://www.youtube.com/watch?v=r7JqcC2h5zk&feature=related

Works for me and is well-suited to architectural subjects.

[Greg Miller]

Quick question, when you work with the luminosity blending mode are you able to see the image in BW ?

Yes. Layer blending modes work for any image where a layer can be created (pretty much everything).

I also use blending mode of Hue for making color balance changes (e.g. a Color Balance layer or Selective Color layer), and a blending mode of Saturation for making saturation changes (e.g. a Saturation layer, or Vibrance layer), or luminosity for a contrast changes (e.g a Curve layer or a Levels layer). I like to isolate hue, saturation, and luminosity changes so they do not affect the other color properties.

[bob carnie]

I would like to see some examples if you don't mind of hue and saturation, blending... I use blending modes quite a bit but usually multiply, screen and soft light then painting in the effect. maybe some screen shots on an image as you work through them if possible as I really never touch hue saturation other than a deliberate bling to a colour..

I use blending modes when I make BW conversions, basically I go into LAB do a spread of colours in curves which separates the colour info, (looks pretty brutal on screen) then back to RGB and Apply Image, picking the right channel is pretty easy *usually the green* and then apply the red to it in moderation and sometimes the blue. then by going through the blending modes I pick the best result , do a simple curve adjustment on the green channel and go to greyscale and get rid of the channels I do not want.

I have found this move to LAB gives the absolute best BW conversion in conjuction with RGB apply image moves. By separating the colours in LAB their seems to be more information to glean in the conversion which give much more tonal separation , specifically in colours that are close to each other and could convert to the same tone which basically increases deadend detail. If that makes any sense to you.

I never use the Adobe BW conversion method .

Yes. Layer blending modes work for any image where a layer can be created (pretty much everything).

I also use blending mode of Hue for making color balance changes (e.g. a Color Balance layer or Selective Color layer), and a blending mode of Saturation for making saturation changes (e.g. a Saturation layer, or Vibrance layer), or luminosity for a contrast changes (e.g a Curve layer or a Levels layer). I like to isolate hue, saturation, and luminosity changes so they do not affect the other color properties.

[Greg Miller]

I primarily only use Hue and Saturation blending modes for color work (not for B&W).

But here is an example where I wanted to cool the shadows and warm up the highlights a tad. Since this objective was about Hue (and not saturation or Luminosity), I set the blending mode to Hue. I rarely increase saturation, because I generally add at least a tad of contract and contrast increases perceived saturation. So almost 100% of the time I use a saturation layer it is to decrease saturation. For contrast, I almost always use a Curves layer with blending mode set to Luminosity - increasing contrast with the blending mode set to normal almost always increases saturation and changes hue, and I want to avoid that.

91892

[Brian Ellis]

I use PKSharpen, usually Capture and Output only, Creative on occasion. When I use Creative I mask the areas I don't want to sharpen.

[Greg Miller]

Here's another sample from the same image. You can see that a very modest curve layer was applied to increase contrast and increase overall luminosity. The Layer Blend Mode is set to Luminosity. Even with this subtle curve, there is a very noticeable difference between Normal and Luminosity blend modes. Leaving blend mode set to normal causes a noticeable shift in both Hue and Saturation. Using a blend mode of Luminosity causes the curve to leave hue and saturation unchanged.

I prefer not to use LAB mode with color images. Staying in RGB gives me 3 color channels to work with (instead of 2), I can still isolate the L channel with blend mode luminosity, and I avoid any damage done by converting to LAB and back.

91923

[Greg Miller]

And in case Drew is reading this, a fairly strong wind was blowing and I was still able to stitch multiple images. ;)

[Greg Miller]

To answer the OP's question, I scan at maximum usable resolution (the scanners highest usable resolution) and work on that scan as the master. I only re-size when I need to output to print or web. I sharpen using Photo Kit Sharpener 2 (like Brian) which will be different for each output size and paper type (matte different than lustre or glossy).

If for some reason I use Phtooshop's sharpen tools, that occurs on a duplicate pixel layer with the layer's blending mode set to Luminosity (like PaulR). This keeps sharpening from affecting hue or saturation, an also allows for masking out areas that should bnot be sharpened.

[paulr]

Here's (http://www.ronbigelow.com/articles/blend-if/blend-if.htm) a tutorial on the "blend if" sliders, which can be used to keep the sharpening out of the brightest and darkest tones. Don't worry if this seems confusing ... it gets intuitive after you play a little.

Here's a screen shot of the blend settings from one of my sharpening layers. this is pretty typical. If you turn off the visibility of the underlying layer while you play with the sliders, you can get a better idea of what's going on.

91932

[Nathan Potter]

Enough guys. It's enough to make a long time Ilfochrome printer weep. ;)

Nate Potter, Austin TX.

[paulr]

Enough guys. It's enough to make a long time Ilfochrome printer weep. ;)

Nate Potter, Austin TX.

What do you mean? Digital's easy ... you just put the thing in the machine and push the button :)

[Nathan Potter]

Haaa! I'm sure we all wished that when we started. A lot of good suggestions here though.

Nate Potter, Austin TX.

[Preston]

Thanks for the info and link to the 'blend if' tutorial, Paul! Good stuff!

I also looked at the sharpening article that was referenced there: good stuff in that one, as well.

--P

[Jim Andrada]

Oh yes - just push the button - any button. And don't say that there's no "any" button on your keyboard!

The big difference between digital and the old original Kodak is that with digital there's nobody for us to "leave the rest" to after we push the button.

[bob carnie]

Thanks Greg

Good points, I usually go to LAB for contrast and the L channel, so as you point out not to increase saturation and changing hue. I will give this a try in RGB and Compare for myself.

I primarily only use Hue and Saturation blending modes for color work (not for B&W).

But here is an example where I wanted to cool the shadows and warm up the highlights a tad. Since this objective was about Hue (and not saturation or Luminosity), I set the blending mode to Hue. I rarely increase saturation, because I generally add at least a tad of contract and contrast increases perceived saturation. So almost 100% of the time I use a saturation layer it is to decrease saturation. For contrast, I almost always use a Curves layer with blending mode set to Luminosity - increasing contrast with the blending mode set to normal almost always increases saturation and changes hue, and I want to avoid that.

91892

[bob carnie]

I never have seen andy damage by convertint to LAB and back.
Using curves in LAB I can set nuetral points and adjust curve on each colour to very subtle points and not have the neutrals shift , also getting rid of unwanted casts is a piece of cake, by setting points on the offending colour and locking the curve . Once locked raising or lowering the point to remove the cast is easy.

All in all I think there are many ways of editing , and each has their preference, I will try using more blending settings in RGB now. thanks

Here's another sample from the same image. You can see that a very modest curve layer was applied to increase contrast and increase overall luminosity. The Layer Blend Mode is set to Luminosity. Even with this subtle curve, there is a very noticeable difference between Normal and Luminosity blend modes. Leaving blend mode set to normal causes a noticeable shift in both Hue and Saturation. Using a blend mode of Luminosity causes the curve to leave hue and saturation unchanged.

I prefer not to use LAB mode with color images. Staying in RGB gives me 3 color channels to work with (instead of 2), I can still isolate the L channel with blend mode luminosity, and I avoid any damage done by converting to LAB and back.

91923

[Ed Richards]

I use Qimage rather than doing output sharpening. I sharpen the Image so it is correct at 100%, then Qimage optimizes it for the print size. It also manages the printer. A remarkable program, and cheap, compared to a RIP.

[bob carnie]

Thanks for that, I have never been able to get my head wrapped around the blend-if feature of photoshop, with Margulis he would speed through this area to colour correct and my fingers were no match for my head and this blend if never caught hold.

Do you use blend if for other reasons than sharpening selections??

Here's (http://www.ronbigelow.com/articles/blend-if/blend-if.htm) a tutorial on the "blend if" sliders, which can be used to keep the sharpening out of the brightest and darkest tones. Don't worry if this seems confusing ... it gets intuitive after you play a little.

Here's a screen shot of the blend settings from one of my sharpening layers. this is pretty typical. If you turn off the visibility of the underlying layer while you play with the sliders, you can get a better idea of what's going on.

91932

[bob carnie]

Sorry to hi jack thread with questions off topic , I may open an Apply Image, and a Blend If Thread if some here are interested.

[SergeiR]

I am but a simple man.. So i do use frequency separation sharpening, and also Nik Sharpener before printing .

[paulr]

Thanks for that, I have never been able to get my head wrapped around the blend-if feature of photoshop, with Margulis he would speed through this area to colour correct and my fingers were no match for my head and this blend if never caught hold.

Do you use blend if for other reasons than sharpening selections??

Honestly, I still find the relationship between the top and bottom sliders to be a bit confusing. Tutorials I've seen actually contradict each other in subtle ways. My method is based a bit on incomplete understanding, a bit on trial and error, a bit on copying Bruse Fraser, and the rest on habit.

In photogrphy I've so far only found the feature useful on the sharpening layer. I've seen people use it for other things, like manually done HDR type blending.

In design and print/web production, which involves a lot more shenanigans, I've used it for other things. For example, it's great if you want to set type and make it look like it's actually printed on a cloth book cover or fabric or brick.

[bob carnie]

Yes very tricky but I think quite powerful, you are the first person I have seen here using it so I am pretty impressed.
Watching Margulis use it you would think it should be in day to day editing , but I have not figured applications in my workflow for it ... yet.


Honestly, I still find the relationship between the top and bottom sliders to be a bit confusing. Tutorials I've seen actually contradict each other in subtle ways. My method is based a bit on incomplete understanding, a bit on trial and error, a bit on copying Bruse Fraser, and the rest on habit.

In photogrphy I've so far only found the feature useful on the sharpening layer. I've seen people use it for other things, like manually done HDR type blending.

In design and print/web production, which involves a lot more shenanigans, I've used it for other things. For example, it's great if you want to set type and make it look like it's actually printed on a cloth book cover or fabric or brick.

[Brian Ellis]

Thanks for that, I have never been able to get my head wrapped around the blend-if feature of photoshop, with Margulis he would speed through this area to colour correct and my fingers were no match for my head and this blend if never caught hold.

Do you use blend if for other reasons than sharpening selections??

If it's still there there's a good explanation of Blend If and another use for it in the Lynda.com tutorial "CS5 Essentials" by Michael Ninness. It's a fairly easy tool to use for a variety of purposes once you get past the non-intuitive name and use it a few times. Lynda.com offers a 7-day free trial.