darkhorse120
Member
....ok, so no unsharp mask applied at scanning...as I want to be in control...what is a 'good' way to sharpen the image in P.Shop? Channels?
sharpening post scanning
- Thread starter darkhorse120
- Start date
- Latest activity Latest activity:
- Replies 15
- Views 14K
ZorkiKat
ЗоркийК&
- Local time
- 7:02 AM
- Joined
- Jan 26, 2006
- Messages
- 2,070
Unsharp mask in the older versions of PS, Smart Sharpen with the newer versions. Amount = 100; Radius = 1; Remove lens blur.
Sharpen only at the final output size- so if you resize the original (as you would if you want to upload it to the net or email it), apply sharpening only after resizing, just before saving as an optimised ('save for web/devices') file.
The smart sharpen/unsharp mask are good ways to sharpen, why not use them?
A lot of sharpness is lost during the scanning process due to the glass/air/film surfaces which the image must pass through. Post sharpening can restore this.
Matus
Well-known
After I have done all other adjustments - including resizing, I usually do 2 stage sharpening (PS CS3):
1) Smart sharpen, radius 0.3, value 50 - 100%, lens blur
- this sharpens small details nicely
- small radius does not sharpen grain so much as radius of 1.0 does.
- value more than 100 or even larger radius may be necessary if the scan is really unsharp - like direct 2400 spi scan from a flat bed scanner. If I have resized the scan to smaller effective dpi - less sharpening seems necessary.
2) Unsharp mask, radius 10.0 - 20.0, value 5 - 10.
- adds more contrast on large scales
- If necessary I my apply two steps with different radii (around 5 and around 15)
1) Smart sharpen, radius 0.3, value 50 - 100%, lens blur
- this sharpens small details nicely
- small radius does not sharpen grain so much as radius of 1.0 does.
- value more than 100 or even larger radius may be necessary if the scan is really unsharp - like direct 2400 spi scan from a flat bed scanner. If I have resized the scan to smaller effective dpi - less sharpening seems necessary.
2) Unsharp mask, radius 10.0 - 20.0, value 5 - 10.
- adds more contrast on large scales
- If necessary I my apply two steps with different radii (around 5 and around 15)
j.scooter
Veteran
Subscribed!
This is the achilles heel of my workflow.
This is the achilles heel of my workflow.
Tim Gray
Well-known
I personally use Unsharp mask. I use it a couple times depending on what I want. Personally, I don't go for the really super sharp look I often see on the DSLR forums; it's all pretty subtle.
1. Very small radius (1 px or less) and high percentage if you want to bring out the grain.
2. Large radius (10-20 px depending on the image size) and percentage up to 500% 😀 No really, do it, then adjust the radius until you find the right details that you want to sharpen (ex. major features and not skin blemishes). Then pull the percentage down until it looks good. Usually around 5-15% for me. You can do this at an even larger scale and lower percentage in a second run if you like.
Number 2 can really bring a lot of clarity to your photo and cut out the haze. Even a subtle setting does a lot; toggle back and forth by turning the preview on and off to see the change. It can be addicting - watch out. I personally back off if I start to see visible sharpening halos.
After resizing, it's usually useful to do another pass of sharpening to tighten things up a bit. You could tailor this step for printing or for screen.
Another useful technique for color is to convert your image into Lab space, go to the L channel, and apply your sharpening there. You can also do this by staying in RGB space, duplicate your base layer, turn it's blending mode to Luminosity, and apply sharpening on that layer. If you do it the second way, you can add in a layer mask and selectively sharpen areas.
1. Very small radius (1 px or less) and high percentage if you want to bring out the grain.
2. Large radius (10-20 px depending on the image size) and percentage up to 500% 😀 No really, do it, then adjust the radius until you find the right details that you want to sharpen (ex. major features and not skin blemishes). Then pull the percentage down until it looks good. Usually around 5-15% for me. You can do this at an even larger scale and lower percentage in a second run if you like.
Number 2 can really bring a lot of clarity to your photo and cut out the haze. Even a subtle setting does a lot; toggle back and forth by turning the preview on and off to see the change. It can be addicting - watch out. I personally back off if I start to see visible sharpening halos.
After resizing, it's usually useful to do another pass of sharpening to tighten things up a bit. You could tailor this step for printing or for screen.
Another useful technique for color is to convert your image into Lab space, go to the L channel, and apply your sharpening there. You can also do this by staying in RGB space, duplicate your base layer, turn it's blending mode to Luminosity, and apply sharpening on that layer. If you do it the second way, you can add in a layer mask and selectively sharpen areas.
wintoid
Back to film
After sharpening my scans for a few years, I stopped. Now I look back on my old scans and think they look rubbish by comparison. I don't feel that scanned film needs sharpening, at least not out of my Minolta Multi Pro, nor the Nikon 5000 I had before it.
Jamie123
Veteran
For medium format scans on my Nikon 9000 for web use I usually don't sharpen at all.
Matus
Well-known
Agreed - scan from a sharp, well exposed film made with film scanner (and not flat-bed) requires little sharpening. Slight sharpening is to me more about getting right the local contrast the way I like it.
Too little sharpening results in the worse case to slightly soft images or prints. On the other hand to much make for images that are just horrible and hurt your eyes ...
One my own example of what I already consider over sharpened image (link goes to full size) :
.. I will have to redo it ..
I have just received a few scans from slides taken by Mamiya 6 on Provia 400X made with Imacon X5 @ 3200 spi. Incomparable to what my Microtek F1 can deliver...
Too little sharpening results in the worse case to slightly soft images or prints. On the other hand to much make for images that are just horrible and hurt your eyes ...
One my own example of what I already consider over sharpened image (link goes to full size) :
.. I will have to redo it ..
I have just received a few scans from slides taken by Mamiya 6 on Provia 400X made with Imacon X5 @ 3200 spi. Incomparable to what my Microtek F1 can deliver...
Gray Fox
Well-known
Matus:
I don't know if it is over sharpened or just a bit too contrasty. I'd suggest reducing the constrast a bit and see what you think. Interesting found still life, though.
I don't know if it is over sharpened or just a bit too contrasty. I'd suggest reducing the constrast a bit and see what you think. Interesting found still life, though.
DrTebi
Slide Lover
I just wanted to add to this thread, that your resizing algorithm also plays a big role in whether and how to sharpen your image.
Photoshop, as well as many other image editors, will give you different options in the resize dialog on how to resample your image. Options are Bicubic, bilinear etc. It is worth playing around with the different options--I noticed that some work better for landscapes, while other work better for closeups or portraits.
ZorkiKat said to only sharpen after your resized your image to your final size. I usually stick to this method as well, although I have found that it sometimes it makes a difference to use a "high-pass sharpening" on the original before resizing it.
The High-Pass filter works really well in my opinion. There are many different ways to set one up, I have setup an action that basically does this:
1. Duplicate Layer
2. Set current layer to overlay mode
3. High Pass Filter with Radius 2.4px (it's in Filters->Others)
4. Set current layer opacity to 50%
It's worth to try this out, give it a shot. More info on the High Pass sharpening can be found here:
http://www.adobe.com/designcenter/photoshop/articles/phscs2at_advsharpen.html
Photoshop, as well as many other image editors, will give you different options in the resize dialog on how to resample your image. Options are Bicubic, bilinear etc. It is worth playing around with the different options--I noticed that some work better for landscapes, while other work better for closeups or portraits.
ZorkiKat said to only sharpen after your resized your image to your final size. I usually stick to this method as well, although I have found that it sometimes it makes a difference to use a "high-pass sharpening" on the original before resizing it.
The High-Pass filter works really well in my opinion. There are many different ways to set one up, I have setup an action that basically does this:
1. Duplicate Layer
2. Set current layer to overlay mode
3. High Pass Filter with Radius 2.4px (it's in Filters->Others)
4. Set current layer opacity to 50%
It's worth to try this out, give it a shot. More info on the High Pass sharpening can be found here:
http://www.adobe.com/designcenter/photoshop/articles/phscs2at_advsharpen.html
ColSebastianMoran
( IRL Richard Karash )
This is a tough subject. Everyone has a point of view.
The best reference I've found is Bruce Fraser, Real World Sharpening with Adobe Photoshop CS2. That's right, CS2. Still a good reference.
The best reference I've found is Bruce Fraser, Real World Sharpening with Adobe Photoshop CS2. That's right, CS2. Still a good reference.
benlees
Well-known
I also use the high pass sharpening. Works well, if a little more time consuming.
DrTebi
Slide Lover
My version of Photoshop is just "CS"... works just fine for me.
But I also use GIMP a lot, as I primarily work on Ubuntu (and very happy here, too).
In GIMP, I really like the "refocus" sharpening filter. I just did some experimenting, and find the results to be quite similar to the high-pass sharpening. It is probably based on the same principle, I haven't investigated further.
@benlees: It does not have to be more time consuming if you just setup an action for the high-pass filtering. Then it's just one click.
But I also use GIMP a lot, as I primarily work on Ubuntu (and very happy here, too).
In GIMP, I really like the "refocus" sharpening filter. I just did some experimenting, and find the results to be quite similar to the high-pass sharpening. It is probably based on the same principle, I haven't investigated further.
@benlees: It does not have to be more time consuming if you just setup an action for the high-pass filtering. Then it's just one click.
J-P
Member
Dr.Tebi - That is also my favoured technique, I've been doing it for years (but only for printed stuff). It allows for a less destructive workflow and also seems to get more pleasing results. You get a nicer preview of the level of sharpening with the high pass mask layer (both previewing the radius whilst applying, and whilst overlayed) - and more control as you control the overlay opacity.
Another advantage - you can get rid of those horrible sharpening halo's that are generated on the skyline by simply masking it out (or indeed, for localised sharpening based on a gradient mask tabbed into your high pass layer).
This method also works in pretty much any version of Photoshop (to my knowledge)... as well as GIMP etc.
The level of desired sharpness surely varies for on-screen and for print (C-type? Electronic? Inkjet?), I know for large C-type prints I've read of people intentionally over-sharpening (with high pass on overlay) - then making a test print and compensating accordingly via opacity. I don't have enough cash or access to such printing equipment to test for real though.
My process:
1) scan image greyscale 16bit TIFF
2) check histogram in levels and adjust accordingly (set black/mid/white)
3) remove any dust/scratches using history brush / clone stamp and edit out any undesirable elements
4) final levels/adjustments/cropping
5) duplicate layer and apply high-pass (layer mode to overlay) - at this point localised sharpening can be done by adding a mask to the overlayed high-pass layer.
6) border if required and prep for print
Another advantage - you can get rid of those horrible sharpening halo's that are generated on the skyline by simply masking it out (or indeed, for localised sharpening based on a gradient mask tabbed into your high pass layer).
This method also works in pretty much any version of Photoshop (to my knowledge)... as well as GIMP etc.
The level of desired sharpness surely varies for on-screen and for print (C-type? Electronic? Inkjet?), I know for large C-type prints I've read of people intentionally over-sharpening (with high pass on overlay) - then making a test print and compensating accordingly via opacity. I don't have enough cash or access to such printing equipment to test for real though.
My process:
1) scan image greyscale 16bit TIFF
2) check histogram in levels and adjust accordingly (set black/mid/white)
3) remove any dust/scratches using history brush / clone stamp and edit out any undesirable elements
4) final levels/adjustments/cropping
5) duplicate layer and apply high-pass (layer mode to overlay) - at this point localised sharpening can be done by adding a mask to the overlayed high-pass layer.
6) border if required and prep for print
DrTebi
Slide Lover
I admit I don't do a lot of printing either. I do however create a second copy of all my scans for screen viewing, which is where I also apply the discussed sharpening techniques.
I believe there are good reasons for creating a second copy for screen presentation:
1. Use the sophisticated resizing algorithm of your image editor to resize the image to your screen, as opposed to having your image viewer do that for you. Any image viewer cares about speed, rather than quality. I have found that most image viewers also "smooth" an image after adjusting it to fit the screen--which will make your images appear less sharp.
2. Your images will open faster. A 30 MB tif scan will take a lot longer to open than an optimized 2 MB JPG.
3. Uploading optimized images to flickr and the like will also be much faster.
4. An optimized JPG also allows me to add exif and/or ITPC data to the image. I have written a very simple application that allows me to add camera, lens, film type and similar information to images. Most of my flickr images have this data, which is a nice way to remember and share your analog camera data.
If you don't believe me... just try it out. Figure out your screen size (e.g. 1920x1080) and resize your scan to that size, then sharpen it. Save this file separately and open this one and the scan in your image viewer--I am sure you will notice the difference 🙂
I believe there are good reasons for creating a second copy for screen presentation:
1. Use the sophisticated resizing algorithm of your image editor to resize the image to your screen, as opposed to having your image viewer do that for you. Any image viewer cares about speed, rather than quality. I have found that most image viewers also "smooth" an image after adjusting it to fit the screen--which will make your images appear less sharp.
2. Your images will open faster. A 30 MB tif scan will take a lot longer to open than an optimized 2 MB JPG.
3. Uploading optimized images to flickr and the like will also be much faster.
4. An optimized JPG also allows me to add exif and/or ITPC data to the image. I have written a very simple application that allows me to add camera, lens, film type and similar information to images. Most of my flickr images have this data, which is a nice way to remember and share your analog camera data.
If you don't believe me... just try it out. Figure out your screen size (e.g. 1920x1080) and resize your scan to that size, then sharpen it. Save this file separately and open this one and the scan in your image viewer--I am sure you will notice the difference 🙂
Ronald M
Veteran
No different from digi camera files. And you do it in stages same as digi camera files.
Ideally, no sharpening until you do noise reduction even if you need to back to ACR with a smart object.
Onen Tiff in ACR, adjust, no sharpen, do the NR at this stage. Open in Photoshopas smart object. Move it back to ACR and do capture sharpen.
Alternative is to do no NR or sharpening in ACR, just basic adjustments. Open in PS. Run the NR of your choice. Then do Capture sharpening, USM or Smart Sharpen.
25/40 amount depending on file size, radius .7, threshold 1.
Resize to final size. Now you sharpen for output and it depends on printing medium, print size, or web. Web would be .5 to .7 radius Amount 75 to 100. Look at the image at 100% and 50% to be sure you are no making halos. If you see halos, reduce amount.
Read "Real World Sharpening" by Frazer & Schewe. You will learn a lot.
Ideally, no sharpening until you do noise reduction even if you need to back to ACR with a smart object.
Onen Tiff in ACR, adjust, no sharpen, do the NR at this stage. Open in Photoshopas smart object. Move it back to ACR and do capture sharpen.
Alternative is to do no NR or sharpening in ACR, just basic adjustments. Open in PS. Run the NR of your choice. Then do Capture sharpening, USM or Smart Sharpen.
25/40 amount depending on file size, radius .7, threshold 1.
Resize to final size. Now you sharpen for output and it depends on printing medium, print size, or web. Web would be .5 to .7 radius Amount 75 to 100. Look at the image at 100% and 50% to be sure you are no making halos. If you see halos, reduce amount.
Read "Real World Sharpening" by Frazer & Schewe. You will learn a lot.
Similar threads
