Maybe everyone on the planet knows about this...it was new to me and I will definitely be using this technique for sharpening from now on. It just provides a look that no other sharpening does.
It just provides a look that no other sharpening does.
But the issue is, visually sharpening doesn’t work so well (sharpening is specific to the image, its size, its final output etc). The disconnect between a low rez display (at differing zoom ratio’s) and a printer is just one reason why.
The LR approach, separating all this (capture and output), and doing so based on what you ask for (in terms of a print size) is so much more effective, non destructive, flexible and based on this old, ground breaking idea of sharpening by the late Bruce Fraser:
"High-Pass" sharpening can be an extremely effective method when it's used properly.
I prefer to do a 3-step sharpening and I do a very light capture sharpen in ACR. Once I'm in Photoshop I choose a sharpening method based on the subject and the goals of the project. If I choose high-pass, the high-pass sharpening method I use is based on:
I would use this in the "creative stage" of sharpening.
I, too, like high pass sharpening. I would add a couple of suggestions to the tute. First, convert the filter layer to a smart object, so that you can go back and tweak the radius of the high pass filter easily. I sometimes find it difficult to judge what the final effect is going to be by looking at the gray-scale preview when applying the filter. The tute does not mention this directly, but I often get the best results using vivid light blending mode--it's easy to cycle through the different blending modes and choose the most appropriate one. I set up an action to go through the whole process, pausing to set radius, blending mode and opacity, which makes it quick and easy.
It had never occurred to me to invert the high pass layer, so that's something I'll have to try. I don't think it's an effect I would be likely to use very often, but I'm curious to see what it will do. Thanks for posting the link, Ric.
I, too, like high pass sharpening. I would add a couple of suggestions to the tute. First, convert the filter layer to a smart object, so that you can go back and tweak the radius of the high pass filter easily. I sometimes find it difficult to judge what the final effect is going to be by looking at the gray-scale preview when applying the filter. The tute does not mention this directly, but I often get the best results using vivid light blending mode--it's easy to cycle through the different blending modes and choose the most appropriate one. I set up an action to go through the whole process, pausing to set radius, blending mode and opacity, which makes it quick and easy.
It had never occurred to me to invert the high pass layer, so that's something I'll have to try. I don't think it's an effect I would be likely to use very often, but I'm curious to see what it will do. Thanks for posting the link, Ric.
Thanks for the info, Richard...this thread is now a bookmark.
Perhaps not "everyone" Ric, however if one spends a short time reading about various sharpening approaches, High Pass filtering is mentioned by a lot of people on many, many different occasions! Reading a bit further, one may stumble over other approaches, such as "Band Pass" or "DoG".
Many of these issues may be new to GUI based users of software such as Photoshop. It is probably 101 level stuff for students of image processing or signal processing (even audio processing, in some ways images and audio have related methods of processing).
Although the following link is more complex than the link in your OP, I think it does a good job of going into some of the deeper details behind HP:
One can get the same result with the USM filter. Further, the HP filter really is nothing special in itself - one can manually reproduce the effect using the Apply Image command or with the Calculations command.
HP filtering is essentially just subtracting a Gaussian blurred dupe of the original image from the original image (there are other things going on)...one can do "better" than the default HP filter by creating their own version using the Apply Image command. What do I mean by "better" - is this not subjective? Less artifacts and junk, as the HP filter does not divide the scale of the subtraction by 2.
This is the basic approach of Unsharp Mask, High Pass and other similar approaches. A blurred copy of the original image is subtracted from the original image to isolate edge detail. This is the difference between the original edges and the blurred edges. These "difference" edges are then amplified and added back in to the original image - creating greater edge contrast and an image that appears to be "sharper".
Then there are other approaches, rather than using the difference between the original and a blurred image - one can find the difference between two blurred images (using different sized blurs in each image). This is known as "DoG" or Difference of Gaussians.
Some of these issues are explored in an old action that I made back in the days of Photoshop 5, which explores the "custom" filter in Photoshop:
Perhaps not "everyone" Ric, however if one spends a short time reading about various sharpening approaches, High Pass filtering is mentioned by a lot of people on many, many different occasions! Reading a bit further, one may stumble over other approaches, such as "Band Pass" or "DoG".
Many of these issues may be new to GUI based users of software such as Photoshop. It is probably 101 level stuff for students of image processing or signal processing (even audio processing, in some ways images and audio have related methods of processing).
Although the following link is more complex than the link in your OP, I think it does a good job of going into some of the deeper details behind HP:
One can get the same result with the USM filter. Further, the HP filter really is nothing special in itself - one can manually reproduce the effect using the Apply Image command or with the Calculations command.
HP filtering is essentially just subtracting a Gaussian blurred dupe of the original image from the original image (there are other things going on)...one can do "better" than the default HP filter by creating their own version using the Apply Image command. What do I mean by "better" - is this not subjective? Less artifacts and junk, as the HP filter does not divide the scale of the subtraction by 2.
This is the basic approach of Unsharp Mask, High Pass and other similar approaches. A blurred copy of the original image is subtracted from the original image to isolate edge detail. This is the difference between the original edges and the blurred edges. These "difference" edges are then amplified and added back in to the original image - creating greater edge contrast and an image that appears to be "sharper".
Then there are other approaches, rather than using the difference between the original and a blurred image - one can find the difference between two blurred images (using different sized blurs in each image). This is known as "DoG" or Difference of Gaussians.
Some of these issues are explored in an old action that I made back in the days of Photoshop 5, which explores the "custom" filter in Photoshop:
The action will create a new doc from a source image. It will then create a High Pass filtered layer for reference. Then the Apply Image command is used to exactly recreate the HP filter effect. This is done with subtraction of a Gaussian blurred copy of the image, using a Scale of 1 and an Offset to 128 (mid gray).
One can compare the effects of the HP filter and the Apply Image results to prove that they are both the same.
There is no real benefit to jumpig through Apply Image hoops if one leaves the Scale at 1 - the HP filter would be simpler and faster.
If one uses a Scale of 2, then I would consider the extra hoop jumping worth while if one is going to use HP filtering. With actions, one can simplify the hoop jumping so that the Apply Image HP effect is just as easy
to use as the default High Pass.
One can experiment with changing the "Scale" value from 1 to 2. Changing the scale value to 2 will allow one to use a stronger contrast blending mode when layering the effect for sharpening. IMHO this creates a better HP effect than the standard HP filter.
It is probably best to explore the results and or the steps in this action. Please let me know if I can help answer any questions that you may have.
Best. Thread. Ever. — on this subject over on Model Mayhem: "HighPass Sucks (+ solution)" (There's been a flaw since day one in the math used in Photoshop's High Pass filter that harms quality, especially detail in the highlights & shadows, which Stephen's and this thread's methods mitigates.)
Comments
Isn't the sharpening in LR3 equivalent to applying USM to the Lab Lightness channel?
Natural selection is responsible for every living thing that exists.
D3s, D500, D5300, and way more glass than the wife knows about.
But the issue is, visually sharpening doesn’t work so well (sharpening is specific to the image, its size, its final output etc). The disconnect between a low rez display (at differing zoom ratio’s) and a printer is just one reason why.
The LR approach, separating all this (capture and output), and doing so based on what you ask for (in terms of a print size) is so much more effective, non destructive, flexible and based on this old, ground breaking idea of sharpening by the late Bruce Fraser:
http://www.creativepro.com/story/feature/20357.html
Author "Color Management for Photographers"
http://www.digitaldog.net/
Not even close. Its based on PhotoKit Sharpener (again, see http://www.creativepro.com/story/feature/20357.html)
Author "Color Management for Photographers"
http://www.digitaldog.net/
I would use this in the "creative stage" of sharpening.
I prefer to do a 3-step sharpening and I do a very light capture sharpen in ACR. Once I'm in Photoshop I choose a sharpening method based on the subject and the goals of the project. If I choose high-pass, the high-pass sharpening method I use is based on:
http://www.dgrin.com/showthread.php?t=2064 (Thanks Waxy!
I have an action for that process and the action allows for considerable adjustment, based on the subject matter.
Final (output) sharpening is either done using the PS downsampling-and-sharpening, or automated with third party software to both downres and sharpen.
Moderator of the Cameras and Accessories forums
It had never occurred to me to invert the high pass layer, so that's something I'll have to try. I don't think it's an effect I would be likely to use very often, but I'm curious to see what it will do. Thanks for posting the link, Ric.
Action? I'd be interested in seeing that...if possible.
Thanks for the info, Richard...this thread is now a bookmark.
Perhaps not "everyone" Ric, however if one spends a short time reading about various sharpening approaches, High Pass filtering is mentioned by a lot of people on many, many different occasions! Reading a bit further, one may stumble over other approaches, such as "Band Pass" or "DoG".
Many of these issues may be new to GUI based users of software such as Photoshop. It is probably 101 level stuff for students of image processing or signal processing (even audio processing, in some ways images and audio have related methods of processing).
Although the following link is more complex than the link in your OP, I think it does a good job of going into some of the deeper details behind HP:
http://www.gamasutra.com/features/20010523/hajba_01.htm
One can get the same result with the USM filter. Further, the HP filter really is nothing special in itself - one can manually reproduce the effect using the Apply Image command or with the Calculations command.
HP filtering is essentially just subtracting a Gaussian blurred dupe of the original image from the original image (there are other things going on)...one can do "better" than the default HP filter by creating their own version using the Apply Image command. What do I mean by "better" - is this not subjective? Less artifacts and junk, as the HP filter does not divide the scale of the subtraction by 2.
This is the basic approach of Unsharp Mask, High Pass and other similar approaches. A blurred copy of the original image is subtracted from the original image to isolate edge detail. This is the difference between the original edges and the blurred edges. These "difference" edges are then amplified and added back in to the original image - creating greater edge contrast and an image that appears to be "sharper".
http://www.panix.com/~rbean/color/color5.txt (short description of darkroom USM)
Then there are other approaches, rather than using the difference between the original and a blurred image - one can find the difference between two blurred images (using different sized blurs in each image). This is known as "DoG" or Difference of Gaussians.
Some of these issues are explored in an old action that I made back in the days of Photoshop 5, which explores the "custom" filter in Photoshop:
http://members.ozemail.com.au/~binaryfx/convolution.html
Sincerely,
Stephen Marsh
http://binaryfx.customer.netspace.net.au/ (coming soon!)
http://members.ozemail.com.au/~binaryfx/
http://prepression.blogspot.com/
http://members.ozemail.com.au/~binaryfx/
http://prepression.blogspot.com/
Oh what did you say, that is not RB?
Looks like I'll have to post more on the topic of HP filtering tomorrow then!
Stephen Marsh
http://binaryfx.customer.netspace.net.au/ (coming soon!)
http://members.ozemail.com.au/~binaryfx/
http://prepression.blogspot.com/
http://members.ozemail.com.au/~binaryfx/
http://prepression.blogspot.com/
http://binaryfx.customer.netspace.net.au/HP-Exploration.zip
The action will create a new doc from a source image. It will then create a High Pass filtered layer for reference. Then the Apply Image command is used to exactly recreate the HP filter effect. This is done with subtraction of a Gaussian blurred copy of the image, using a Scale of 1 and an Offset to 128 (mid gray).
One can compare the effects of the HP filter and the Apply Image results to prove that they are both the same.
There is no real benefit to jumpig through Apply Image hoops if one leaves the Scale at 1 - the HP filter would be simpler and faster.
If one uses a Scale of 2, then I would consider the extra hoop jumping worth while if one is going to use HP filtering. With actions, one can simplify the hoop jumping so that the Apply Image HP effect is just as easy
to use as the default High Pass.
One can experiment with changing the "Scale" value from 1 to 2. Changing the scale value to 2 will allow one to use a stronger contrast blending mode when layering the effect for sharpening. IMHO this creates a better HP effect than the standard HP filter.
It is probably best to explore the results and or the steps in this action. Please let me know if I can help answer any questions that you may have.
Hope this helps,
Stephen Marsh
http://binaryfx.customer.netspace.net.au/ (coming soon!)
http://members.ozemail.com.au/~binaryfx/
http://prepression.blogspot.com/
http://members.ozemail.com.au/~binaryfx/
http://prepression.blogspot.com/
"HighPass Sucks (+ solution)"
(There's been a flaw since day one in the math used in Photoshop's High Pass filter that harms quality, especially detail in the highlights & shadows, which Stephen's and this thread's methods mitigates.)