Sharpening scans in Lightroom

[Alan Edward Klein]

How would you sharpen past 150 on the LR sharpening slider? When I scan at 4800bpi, I need more sharpening than the 150. It's OK when I scan at 2400. I use LR Version 6.0 purchased, not CC.

 

[bernard_L]

You might try unsharp mask (USM) sharpening. It has two main parameters: sigma (a spatial scale, typically of order one to a few pixels) and amount. The single "amount" parameter on your slider cannot fully substitute for the two USM parameters. I don't have LR, but USM is a basic method, and I would expect it to be present somewhere in the LR menus.

 

[Alan9940]

Could you apply some initial sharpening in the scanner software?

 

[bernard_L]

Could you apply some initial sharpening in the scanner software?

I don't know which scanning software you use. In vuescan (what I use), in the Filter tab, there is a "sharpen" option with just two values: OFF or ON. Like most other processing options, this is best left to post-processing. Depending on circumstances, I may apply USM sharpening in one of three environments:

• ImageMagick: with command line
  magick convert infile -unsharp 0x1.4+1.0+0.02 outfile
  with 0 being shorthand for auto-radius, 1.4 my starting value for sigma (often called radius in other software), 1.0 the amount (equivalent to 100% in other software notation) and 0.02 a plausible value for thershold, to avoid sharpening the noise. With a simple console script, you can process dozens or hundreds of files in one go.
  
• FastStone. Sharpen Tool, USM option. I would start at Amount: 35 (percent??), radius 1.0. Interactive preview.
  
• Picture Window Pro 8.0. Also interactive preview in the sharpen tool. Can create scripts for batch processing.

Each of these is free (as in free beer).

 

[Alan Edward Klein]

I use Epsonscan to scan
I don't sharpen during the scan. Maybe I have to. I don't need tp go higher than 150 1.0 when I sharpened a 2400 bpi scan. It seems I need more when I scan at 4800. I guess it had something to do with the size of the bits. I could try changing the 1.0 to a higher number.

 

[MattKing]

Sharpening artificially lines up the bits that are otherwise broken into non-aligned batches.
A 2400 ppi scan has fewer and larger bits to align, so you see the results more clearly on the screen.
Use of USM that allows you to control both the strength and radius of the effect allows you to customize your result to match what you are working with.

 

[jeffreyg]

Check out Topaz Labs sharpen ai. They have a free trial. If you like it it costs $79.99. I use it when necessary but always try to have a sharp negative as my first choice.

http://www.jeffreyglasser.com/

http://www.sculptureandphotography.com/

 

[Alan Edward Klein]

What LR settings do people use sharpening scans?

 

[grat]

I've become a fan of "frequency separation"-- usually used for portraits, but works well on landscape photography too. Robin Whalley on YouTube has some nice tutorials on it.

It's a nice way to sharpen the details, without boosting the grain too much.

 

[jtk]

Fwiw I think there's a concensus that alleged sharpness stops being "real" beyond 2400...thats just a setting at which point the scanner starts to fake things. If you're after sharpness in prints your application (eg NIK) can tweak with that especially in mind.

 

[grat]

Fwiw I think there's a concensus that alleged sharpness stops being "real" beyond 2400...thats just a setting at which point the scanner starts to fake things. If you're after sharpness in prints your application (eg NIK) can tweak with that especially in mind.

There was a "consensus" that bumblebees have wings that are too small to support their body, that the earth is flat, and that if you drive faster than 30 miles per hour in a car, you'll suffocate because you can't breathe at that speed.

While I consider 2400 PPI to be a reasonable resolution to scan most of my negatives, my scanner will go beyond that a bit, depending on the source. You're unlikely to get much more than 2800 PPI out of a 35mm negative (and you'll struggle with that due to film curve), but there does appear to be useful data in the 4000+ range for medium / large format.

There is some softness, particularly with the Epsons, but it can be recovered.

 

[runswithsizzers]

What LR settings do people use sharpening scans?

Generally speaking, I will choose the following Lightroom "Detail" (Sharpening) settings for digitized 35mm film:
Amount: 20-50
Radius: 1.2-2.4
Detail: Usually left at Lightroom Default of 25, but sometimes reduced if there is a lot of noise or objectionable grain.
Masking: Usually left at Lightroom Default of 0, but sometimes increased if there is a lot of noise or objectionable grain.

I am very sensitive to oversharpening and would prefer to leave my film images a little soft rather than risking oversharpening. If I think a film scan needs additional sharpening, I usually start by increasing the Radius first, rather than the Amount.

My digitizing method has changed over the years. Most recently, I have been using a 16MP Fuji mirrorless camera with resolution of 4896 x 3264 pixels. (lens: Rodenstock APO-Rodagon D 75mm f/4.0). My earliest scans were done on a Minolta Dimage Scan Elite F-2900 film scanner (2820 ppi optical resolution), and Vuescan. I believe I used similar settings when sharpening the Minolta scans, but some of those were sharpened in Photoshop rather than Lightroom, so I can't recall what exact settings were used.

My workflow for b&w usually starts in Photoshop rather than Lightroom. In Photoshop I sometimes use Smart Sharpen or Unsharp mask, but most often I like to use Photoshop's High Pass Filter with a Radius of 0.8-1.8.

If you want to see results, <this SmugMug album> is my most recent - Ektar negatives "scanned" with my Fuji+copy lens. I did not apply more than an Amount = 40 to any of these amd most got an Amount of 22-28 and Radius 1.3-1.4. However! I just discovered Negative Lab Pro plug-in applied a significant amount of sharpening when doing the color inversion. I had the NLP conversion set to Sharpen=Lab which resulted in Lightroom Amounts of 80-83, Radius of 1.1 and Masking of 50. So my Lightroom sharpening was apparently appled on top of what NLP had already applied.

 

[runswithsizzers]

So far, no one has mentioned output sharpening. I don't understand how all this works well enough to discuss it, but I've been told that sharpening applied in Lightroom only applies to the full res image in Lightroom, and if you Export at any other resolution than all that really matters is the output sharpening. Can anyone say for sure if output sharpening is in addition to what is applied using Lightroom's Detail tools - or does output sharpening replace Lightroom sharpening amounts?

 

[jtk]

So far, no one has mentioned output sharpening. I don't understand how all this works well enough to discuss it, but I've been told that sharpening applied in Lightroom only applies to the full res image in Lightroom, and if you Export at any other resolution than all that really matters is the output sharpening. Can anyone say for sure if output sharpening is in addition to what is applied using Lightroom's Detail tools - or does output sharpening replace Lightroom sharpening amounts?

See my post above. If you want to maximize results without creating artifacts you need to specify a specific intended OUTPUT, eg which printer, which sort of online display.

I don't think there is any totally abstract, unspecified "output".

 

[ced]

I would first determine the final use of the image & how it will be printed before any USM is applied.
If the image is just for archival leave it as it comes from the scanner without USM @ 2400 ppi or 4000+ppi the image is at 1:1 but as you reduce the resolution for the necessary print output resolution the size will change unless you alter the constraints or not. Then you can apply your wanted USM params. When the image is to be used just on the web the settings will be very low.

 

[Alan Edward Klein]

Generally speaking, I will choose the following Lightroom "Detail" (Sharpening) settings for digitized 35mm film:
Amount: 20-50
Radius: 1.2-2.4
Detail: Usually left at Lightroom Default of 25, but sometimes reduced if there is a lot of noise or objectionable grain.
Masking: Usually left at Lightroom Default of 0, but sometimes increased if there is a lot of noise or objectionable grain.

I am very sensitive to oversharpening and would prefer to leave my film images a little soft rather than risking oversharpening. If I think a film scan needs additional sharpening, I usually start by increasing the Radius first, rather than the Amount.

My digitizing method has changed over the years. Most recently, I have been using a 16MP Fuji mirrorless camera with resolution of 4896 x 3264 pixels. (lens: Rodenstock APO-Rodagon D 75mm f/4.0). My earliest scans were done on a Minolta Dimage Scan Elite F-2900 film scanner (2820 ppi optical resolution), and Vuescan. I believe I used similar settings when sharpening the Minolta scans, but some of those were sharpened in Photoshop rather than Lightroom, so I can't recall what exact settings were used.

My workflow for b&w usually starts in Photoshop rather than Lightroom. In Photoshop I sometimes use Smart Sharpen or Unsharp mask, but most often I like to use Photoshop's High Pass Filter with a Radius of 0.8-1.8.

If you want to see results, <this SmugMug album> is my most recent - Ektar negatives "scanned" with my Fuji+copy lens. I did not apply more than an Amount = 40 to any of these amd most got an Amount of 22-28 and Radius 1.3-1.4. However! I just discovered Negative Lab Pro plug-in applied a significant amount of sharpening when doing the color inversion. I had the NLP conversion set to Sharpen=Lab which resulted in Lightroom Amounts of 80-83, Radius of 1.1 and Masking of 50. So my Lightroom sharpening was apparently appled on top of what NLP had already applied.

Does that include any sharpening during the scan?

 

[runswithsizzers]

Does that include any sharpening during the scan?

I use those approximate settings (the ones at top of my first post), with several different workflows. When I was scanning with the Minolta film scanner and Vuescan, no, I did not set Vuescan to apply any sharpening. My assumption was that I should probably be able to control the results better using Adobe's sharpening tools.

I have also found those setting to be about right with some "basic scans" I get back with processed C-41 negs from one lab I have used. (Some other labs oversharpen their scans to a ridiculous degree, and the results are unacceptable.)

Another workflow is to copy the film with my digial camera. I use RAW files from the camera, so there is no in-camera sharpening. When I open the RAW file in Lightroom Classic, a default amount of sharpening is applied to RAW files, Amount: 40, Radius: 1.0, Detail: 25, Masking: 0. As mentioned above, for copies of film, I often increase the Radius from the Lr default of 1.0 to 1.2-2.4 (sometimes a little more).

For b&w negs and slides (which do not need to be inverted with NLP), I open the RAW files from my Fuji digital camera in Photoshop rather than Lightroom. I like Photoshop's dust removal tools much better than Lightroom's. Of couse that means passing the RAW file through ACR, where the Default sharpening settings are the same as in Lightroom, and where I bump up the Radius, same as for Lightroom.

 

[runswithsizzers]

See my post above. If you want to maximize results without creating artifacts you need to specify a specific intended OUTPUT, eg which printer, which sort of online display.

I don't think there is any totally abstract, unspecified "output".

I agree, output sharpening is best done for a specific resolution and for the intended use of the file. But output sharpening is not usually applied until/unless a file is Exported from Lightroom, right? The Lightroom Export dialogs allow me to specify if the output sharpening is to be done for Screen, Matt Paper, or Glossy Paper, and if I want the amount of sharpening to be Low, Standard, or High. (I don't print directly from Lightroom, so I don't know if the Lightroom print module does any additional sharpening at the time of printing.)

What I was wondering is this, how does any sharpening previously done on a file in Lightroom affect the results of subsequent output sharpening done at the time of export? I assume the two sharpening steps are additive(?) - but is sharpening done in Lightroom's Detail panel likely to improve output sharpening, or can this pre-sharpening adversely affect the quality of the output sharpening in some way? In a perfect world, I would make the image look like I want it on my screen, and any output sharpening would be just enough to restore any loss of sharpness due to the export process, so the exported file will look just like it did on my screen before I exported it.

 

[Robert Maxey]

There was a "consensus" that bumblebees have wings that are too small to support their body, that the earth is flat, and that if you drive faster than 30 miles per hour in a car, you'll suffocate because you can't breathe at that speed.

Consensus or not, can't it be proven with a few tests? I do not know, but how good a SW package does this or that or not really much at all, after some limit could be proven with a few tests, keep accurate records and make a print. I must say that I've seen too many over sharpened images and do not get me started on HDR.

Bob

 

[Kodachromeguy]

Does that include any sharpening during the scan?

Adrian, who often posts here, warned me to NOT sharpen during the scan process. Do the sharpening afterwards with one of many specialty programs. Do the scan at the optical resolution of the scanner, if you can find that statistic.

 

[grat]

Consensus or not, can't it be proven with a few tests? I do not know, but how good a SW package does this or that or not really much at all, after some limit could be proven with a few tests, keep accurate records and make a print. I must say that I've seen too many over sharpened images and do not get me started on HDR.

And that's where the arguments start. For example, one scanning site plopped the USAF 1951 target down on the bed of the scanner (orientation unknown), and scanned it. On the Epson v7xx/8xx scanners, this means you're using the "low" resolution sensor, and automatically limited to 4800 PPI. They evaluated a blurry, out-of-focus scan to determine which line pair was still distinguishable, and announced the scanner is only capable of 2300 PPI. 2600 if you use different software-- which to me, means they're testing software, not hardware. And since it was blurred, it wasn't a very good test.

Another paper quoted by a particular member, claims to have calculated the MTF, based on an analysis of a similarly bad scan, and declared that the MTF on the Epsons is so bad, that they can only manage 700 PPI. For me, any result (like the calculation that "proved" bumblebees cannot fly) that is disproved by empirical evidence, is highly suspect.

My personal experience with a v800 is that going beyond 3200 PPI on a 35mm negative might be a waste of time, and really, if you scan at 3200, sharpen in post, and downsample to 2400 PPI, you'll produce very nice results.

I've uploaded some samples to https://flickr.com/photos/ufgrat/albums/72157719390119774/with/51236492602/, none of which I claim are particularly spectacular, as they're all on 400 speed film-- The first two are 35mm at 3200 and 6400 ppi respectively, the 3rd is 4x5 at 3200 PPI (and Rollei Infrared 400), and the last image is a 6x6 120 scanned at 2400 (it claims 240, but I'm not sure why). The magnolia is also the oldest scan-- I need to rescan it at higher resolution.

My personal workflow is to scan with SilverFast, with minimal settings-- basically, negafix, histogram equalization if it needs it, and infrared based scratch/dust removal for color negatives. Everything else is left off. Then in Affinity Photo, I correct color balance, remove dust/scratches, see if an RGB curve will help, and finally apply the "frequency separation" sharpening I mentioned in an earlier post.

 

[Alan Edward Klein]

I use those approximate settings (the ones at top of my first post), with several different workflows. When I was scanning with the Minolta film scanner and Vuescan, no, I did not set Vuescan to apply any sharpening. My assumption was that I should probably be able to control the results better using Adobe's sharpening tools.

I have also found those setting to be about right with some "basic scans" I get back with processed C-41 negs from one lab I have used. (Some other labs oversharpen their scans to a ridiculous degree, and the results are unacceptable.)

Another workflow is to copy the film with my digial camera. I use RAW files from the camera, so there is no in-camera sharpening. When I open the RAW file in Lightroom Classic, a default amount of sharpening is applied to RAW files, Amount: 40, Radius: 1.0, Detail: 25, Masking: 0. As mentioned above, for copies of film, I often increase the Radius from the Lr default of 1.0 to 1.2-2.4 (sometimes a little more).

For b&w negs and slides (which do not need to be inverted with NLP), I open the RAW files from my Fuji digital camera in Photoshop rather than Lightroom. I like Photoshop's dust removal tools much better than Lightroom's. Of couse that means passing the RAW file through ACR, where the Default sharpening settings are the same as in Lightroom, and where I bump up the Radius, same as for Lightroom.

Thanks. Your pictures in smugmug are very nice.