Okay, so that's one of those misleading and sensational thread titles I hate so much.
But seriously, I still(! like 99% of PS professionals!) use High Pass, despite Sean Baker's groundbreaking research. Despite Photons 2 "the action man" Pixel's best efforts, I still feel that HP is faster and more intuitive to use. Really, all Sean has pointed out is that it is not an exact inverse of Gaussian Blur. Adobe, to my knowledge has never claimed it was, and it's hardly useless, even if it sucks a [/i]little[/i].
What I actually meant to post is this, and I've never seen/heard anybody use this:
When I use a merged copy for sharpening I always desaturate to avoid funky color fringing in high contrast areas. A while back I stopped to ask why make a layer grayscale when I already have three just sitting there.
So usually for sharpening skin, I'll start with the red channel, since there's less detail in the skin, but still plenty of contrast in eyes etc. Even if I plan to mask the whole layer and paint in the sharpening, starting with red only lets me move a little more quickly (read: sloppily), as I frequently have to work very efficiently just to stay profitable.
Does everybody do this, and not mention it? Is there some glaring reason I shouldn't? (Besides the inherent suckage of High Pass)
Also, if I feel the skin is too smooth, sometimes I'll use the green channel, or even blue.
ezpkns wrote: So usually for sharpening skin, I'll start with the red channel, since there's less detail in the skin, but still plenty of contrast in eyes etc. Even if I plan to mask the whole layer and paint in the sharpening, starting with red only lets me move a little more quickly (read: sloppily), as I frequently have to work very efficiently just to stay profitable.
Does everybody do this, and not mention it? Is there some glaring reason I shouldn't? (Besides the inherent suckage of High Pass)
Also, if I feel the skin is too smooth, sometimes I'll use the green channel, or even blue.
Discuss/complain/educate/bicker.
It's one of the basic tenets of Dan Margulis' teachings, only he does it to the black and or cyan channel in cmyk. He doesn't necessarily advocate doing it on a separate layer and then converting blend mode to luminosity, but that works very well too.
Edit... mostly this applies to lighter skinned people, although it can apply to many things.
with HP I can visually decide on the lower frequency cut of band I like to use...
(showing High to band's Low-cut point, which I usually need)
with frequency separation I can't easily pick where to split the frequency.
(when picking the point separation you have the Low freq band separation visualized, and I find it hard to infer the visualization of higher frequency result that I need)
ShadowLight wrote: with HP I can visually decide on the lower frequency of band I like to use...
(showing High to band's Low-cut point, which I usually need)
with frequency separation I can't easily pick where to split the frequency.
(when picking the point separation you have the Low freq band separation visualized, and I find it hard to infer the visualization of higher frequency result that I need)
Exactly what I mean by more intuitive. Sean and Photon et al have worked wonders to make their way easier to use, but what would be best is if Adobe rewrote High Pass for CS5 to be a direct inverse corollary to G. Blur. I'm sure SB wouldn't complain!
Exactly what I mean by more intuitive. Sean and Photon et al have worked wonders to make their way easier to use, but what would be best is if Adobe rewrote High Pass for CS5 to be a direct inverse corollary to G. Blur. I'm sure SB wouldn't complain!
I'm sure no one would complain. It certainly would simplify things a lot. Then we wouldn't need any of "the action man's" actions.
"the action man"..... Nice one. I do scripts, too. Maybe a new name for me? Scraption Man? Scription Man? Scraction Man?
Of note and included in the thread you're referring to, the High Pass filter is the inverse of the GB function - it simply does not account for contrast addition / range appropriately.
SRB Photo wrote: Of note and included in the thread you're referring to, the High Pass filter is the inverse of the GB function - it simply does not account for contrast addition / range appropriately.
If I take a few days (okay, weeks) off from readingthat thread, it's very daunting to try to catch up. Thank you for clarifying here. I hope you don't get the impression that my post was meant to detract from your research (which I've personally benefitted from greatly, btw).
So that's the challenge for Adobe's PS engineers, then - accounting and adjusting for changes in contrast and range. In both 8 and 16 bits, while I'm sitting on Santa's knee...
Photons 2 Pixels Images wrote: "the action man"..... Nice one. I do scripts, too. Maybe a new name for me? Scraption Man? Scription Man? Scraction Man?
I'd leave the scripts part out. You just can't make scripts sound "cool", even if you name them after snakes...
LeDeux Art wrote: i have no complaints with H.P., ill use it for certain projects and i dont rely on it all the time. I try to get it right in camera as much as i can
You are probably using some form of DSLR in your photography. The filters used over the sensor in those cameras makes post process sharpening a necessity. That sharpening can take many forms, one of them being the prudent use of the high pass filter. There is no "getting it right in camera" to be considered here.
If you are using film, and running that through a scanner, you face the same dilemma. To the point that on a Heidleberg Tango drum scanner, you cannot defeat the sharpening, at some level, it is on all the time.
For some reason, you seem to have taken on the "get it right in camera" mantra, as if you believe you actually do get it right in camera. You post the same statement all the time to the point of nausea. Offering suggestions about subjective issues such as "getting it right in camera", only serve to open your work to the scrutiny of others who may not think you get it right in camera at all. This habit of yours detracts from the issues pertinent to the thread, and does nothing to add to anyones knowledge base. While it may seem like a good way to serve your more basic needs for attention, it actually puts you in a negative light. You might consider these things the next time you feel a need to let us all know how you "get it right in camera".
You are probably using some form of DSLR in your photography. The filters used over the sensor in those cameras makes post process sharpening a necessity. That sharpening can take many forms, one of them being the prudent use of the high pass filter. There is no "getting it right in camera" to be considered here.
If you are using film, and running that through a scanner, you face the same dilemma. To the point that on a Heidleberg Tango drum scanner, you cannot defeat the sharpening, at some level, it is on all the time.
For some reason, you seem to have taken on the "get it right in camera" mantra, as if you believe you actually do get it right in camera. You post the same statement all the time to the point of nausea. Offering suggestions about subjective issues such as "getting it right in camera", only serve to open your work to the scrutiny of others who may not think you get it right in camera at all. This habit of yours detracts from the issues pertinent to the thread, and does nothing to add to anyones knowledge base. While it may seem like a good way to serve your more basic needs for attention, it actually puts you in a negative light. You might consider these things the next time you feel a need to let us all know how you "get it right in camera".
Unless of course you're a JPG only shooter (which I'm not bye the way) in which case the camera's own internal JPG conversion software will add some kind of sharpening in. Some people seem perfectly happy with that.
Justin Foto wrote: Unless of course you're a JPG only shooter (which I'm not bye the way) in which case the camera's own internal JPG conversion software will add some kind of sharpening in. Some people seem perfectly happy with that.
While it all happens at the speed of light, and the application of the sharpening you are talking cannot probably be measured in terms of their point of actuation on a time slider, I assure you it happens sequentially, and is indeed post processing. However, anyone that uses the basic sharpening allowed by choice, in camera, probably deserves what they get when going out to print. My suggestion would be to turn off in camera sharpening, and do a much better job of it in post.
But you bring up a good point about JPG shooters. I'm going to start another thread concerning that.
While it all happens at the speed of light, and the application of the sharpening you are talking cannot probably be measured in terms of their point of actuation on a time slider, I assure you it happens sequentially, and is indeed post processing. However, anyone that uses the basic sharpening allowed by choice, in camera, probably deserves what they get when going out to print. My suggestion would be to turn off in camera sharpening, and do a much better job of it in post.
But you bring up a good point about JPG shooters. I'm going to start another thread concerning that.
I don't understand JPG only shooters. To me, half the fun is in the back-end - what used to be the realm of the darkroom. JPG only is like Ansel Adams dropping film off at a 1 hour photo booth. I personally don't like the way my Camera's handle color rendition or sharpening when processing JPG's. It's not to my taste.
Justin Foto wrote: I don't understand JPG only shooters. To me, half the fun is in the back-end - what used to be the realm of the darkroom. JPG only is like Ansel Adams dropping film off at a 1 hour photo booth. I personally don't like the way my Camera's handle color rendition or sharpening when processing JPG's. It's not to my taste.
While I shoot raw all the time, and with MF digital, you can't not shoot raw, I am perplexed a bit by the notion that raw provides so much more potential than a jpg. Raw files are basically a bucket of pixels with predetermined parameters, that you change by adjusting those parameters. I can do the same to a jpg. I can change the parameters of a raw file to the point at which the file is corrupt, at least to my eyes. I can do the same to a jpg. I've been doing this so long, exposure and color balance are never really an issue, so where is the advantage for me in shooting raw, and exactly how different is a raw file from a jpg. For instance... a blown highlight is blown, whether it be in raw or jpg.
Robert Randall wrote: While I shoot raw all the time, and with MF digital, you can't not shoot raw, I am perplexed a bit by the notion that raw provides so much more potential than a jpg. Raw files are basically a bucket of pixels with predetermined parameters, that you change by adjusting those parameters. I can do the same to a jpg. I can change the parameters of a raw file to the point at which the file is corrupt, at least to my eyes. I can do the same to a jpg. I've been doing this so long, exposure and color balance are never really an issue, so where is the advantage for me in shooting raw, and exactly how different is a raw file from a jpg. For instance... a blown highlight is blown, whether it be in raw or jpg.
I have to disagree with that. For those of us who are still really only learning the art of lighting, I have found that there is much more leeway with blown highlights in RAW that in JPG. It's about dynamic range. A blown highlight in RAW bight show blown in a RAW editor, but it's in the 11th bit of a 12 bit system, the detail is still there. Not so in JPG.
In Lightroom, I can use the exposure brush on what looks like a blown highlight and find the detail is still there and I can rescue a badly lit file. In JPG, on the same file, it's just gone. (I used to shoot RAW+JPG, so I had both files available)
Obviously, as I get better with lighting, I find I do this less and less.
Justin Foto wrote: I have to disagree with that. For those of us who are still really only learning the art of lighting, I have found that there is much more leeway with blown highlights in RAW that in JPG. It's about dynamic range. A blown highlight in RAW bight show blown in a RAW editor, but it's in the 11th bit of a 12 bit system, the detail is still there. Not so in JPG.
In Lightroom, I can use the exposure brush on what looks like a blown highlight and find the detail is still there and I can rescue a badly lit file. In JPG, on the same file, it's just gone. (I used to shoot RAW+JPG, so I had both files available)
Obviously, as I get better with lighting, I find I do this less and less.
I'm not so sure about that. It's been said time and time again, but My understanding is a bit different. Though, I admit, I could very well be wrong and usually I am.
For reference to my way of understanding, I'll give an example.
I want to count apples in an orchard. There are 2 programs I can use for this. One program counts each apple and the other only in multiples of 100. So for each apple I count, I hit a button in the first program and for every 100 I count, I hit a button in the second one also. They each let me know when I hit the proper amount of apples to fit on a truck for delivery...we'll associate this with pure white in RGB (255/255/255). Now, program 1 and program 2 will give me the same amount that I can put on the truck, the only difference is that program 1 will have a more accurate count so if there is a partial delivery, I know exactly how many apples went on it vs program 2 which is rounded to the nearest 100.
12 bit or 16 bit is simply more accurate between the values of the 8 bit "roundings" but I don't believe you can grab more information. Or am I wrong?
I'm not so sure about that. It's been said time and time again, but My understanding is a bit different. Though, I admit, I could very well be wrong and usually I am.
For reference to my way of understanding, I'll give an example.
I want to count apples in an orchard. There are 2 programs I can use for this. One program counts each apple and the other only in multiples of 100. So for each apple I count, I hit a button in the first program and for every 100 I count, I hit a button in the second one also. They each let me know when I hit the proper amount of apples to fit on a truck for delivery...we'll associate this with pure white in RGB (255/255/255). Now, program 1 and program 2 will give me the same amount that I can put on the truck, the only difference is that program 1 will have a more accurate count so if there is a partial delivery, I know exactly how many apples went on it vs program 2 which is rounded to the nearest 100.
12 bit or 16 bit is simply more accurate between the values of the 8 bit "roundings" but I don't believe you can grab more information. Or am I wrong?
You are indeed wrong. You're not counting apples in both cases. You're counting apples in one case and a much smaller fruit in the other case. There's finer detail in RAW - so it can be rescued. That detail just doesn't exist in JPG. Also, even if it could exist - say it's within the 8 bit boundary, the JPG lossy algorithm would blow it away anyway.
Robert Randall wrote: I'm thinking that if a portion of a raw file, or a jpg file, reads at 255, there is no detail, and none to be salvaged. Is that wrong?
No, you're not wrong. If it really is 255, it really has gone. I was talking about a finer level of detail than that. RAW can count between 244 and 255. JPG can't. More to the point, JPG will throw away detail that RAW won't.
i used H.P. on my current avi. she had deep eyes and they were lit well in a small % of the shots. i liked her expression here and used photoshop to compensate the lack of light in her sockets. Thats a good example of when i use it
LeDeux Art wrote: i have no complaints with H.P., ill use it for certain projects and i dont rely on it all the time. I try to get it right in camera as much as i can
Robert Randall wrote: I'm thinking that if a portion of a raw file, or a jpg file, reads at 255, there is no detail, and none to be salvaged. Is that wrong?
You know this so i don't know why I am writing this, but a RAW file is probably 12bits, 14bits, 16 bits or more. Therefore there is more headroom to lighten shadows and reduce highlights down to an 8bit TIFF/JPG than if you start with blown highlights/lowlights in an 8bit JPG.
The values fixed in the 8bit JPG based the exposure setting at the time are a subset of the higher bit exposure values available in the RAW file. Therefore you can quite satisfactorily recover highlights and shadows within reason.
I guess this does let you get it right in camera, if you get it wrong in camera!
RAW files from Nikon and Canon (someone else can look up the others?) are 12 or 14bit.
Blown highlights are blown highlights. But highlights that are not-quite-blown may yet be recoverable from a RAW, whereas the camera's own jpeg converter may unwittingly discard the information that might have made recovery possible.
Although none of this is relevant to the get-it-right-in-camera brigade, who've once again stumbled in en masse to the wrong forum.
ezpkns wrote: I hope you don't get the impression that my post was meant to detract from your research (which I've personally benefitted from greatly, btw).
The title might confuse one, as well the suggestions of intent and 'best courses'.
ezpkns wrote: So that's the challenge for Adobe's PS engineers, then - accounting and adjusting for changes in contrast and range. In both 8 and 16 bits, while I'm sitting on Santa's knee...
The right move, and what's been suggested + submitted is to simply use existing code to halve the image contrast up front, and then use the existing to extract at the selected frequency. As well, an option to 'maximize contrast' for those who want the LCE value could be included to lessen contrast only enough to avoid clipping.
As to your original question, if you're using HP to sharpen, I'm curious why you do so vs. USM?
Robert Randall wrote: I've been doing this so long, exposure and color balance are never really an issue, so where is the advantage for me in shooting raw, and exactly how different is a raw file from a jpg. For instance... a blown highlight is blown, whether it be in raw or jpg.
The two issues you'd likely have [to work around], assuming perfect color and exposure in capture, would be the tone curve applied by the body (vs. one which you control in post), as well compression artifacts if your ADs / CDs are still looking @ 400% for problems with the image. Otherwise, all other things being equal, you'd do as well as in raw.
ezpkns wrote: I hope you don't get the impression that my post was meant to detract from your research (which I've personally benefitted from greatly, btw).
SRB Photo wrote: The title might confuse one, as well the suggestions of intent and 'best courses'.
Not sure I follow you; your own initial post in that thread practically confesses the same thread titling hyperbole. I honestly don't know what you mean regarding "intent" or "best courses". Are you under the impression that I'm positing my method as superior? I doubt that anyone else has read that into it. Your results are irrefutable, using the "difference blend mode litmus test." I hope I didn't give anyone the impression that I think otherwise.
ezpkns wrote: So that's the challenge for Adobe's PS engineers, then - accounting and adjusting for changes in contrast and range. In both 8 and 16 bits, while I'm sitting on Santa's knee...
SRB Photo wrote: The right move, and what's been suggested + submitted is to simply use existing code to halve the image contrast up front, and then use the existing to extract at the selected frequency. As well, an option to 'maximize contrast' for those who want the LCE value could be included to lessen contrast only enough to avoid clipping.
As to your original question, if you're using HP to sharpen, I'm curious why you do so vs. USM?
I presumed you already submitted it.
Why do I use High Pass over USM? Are you testing me, or actually curious? It's better, that's why. Good enough?
If you're implying that if I'm looking for a way to sharpen high-frequency information separately, you have a better way, I think I acknowledged that in my original post. If you're saying that I may as well use the threshold slider in USM, thank you, but I'd rather not. K?
Robert Randall wrote: I'm thinking that if a portion of a raw file, or a jpg file, reads at 255, there is no detail, and none to be salvaged. Is that wrong?
That's just the thing. JPG is 8bits per pixel (0-255) or 256 discrete values
A 12bit RAW file is 4096 discrete values
A 14bit RAW file is 16,384 discrete values
A 16bit RAW file is 65,536 discrete values
HUGE difference
If you divide in you'll see that a 14bit RAW has 64 TIMES the number of possible tonal values per pixel than an 8bit jpeg. Nevermind the compression artifacting and other issues.
ezpkns wrote: Not sure I follow you; your own initial post in that thread practically confesses the same thread titling hyperbole. I honestly don't know what you mean regarding "intent" or "best courses". Are you under the impression that I'm positing my method as superior? I doubt that anyone else has read that into it. Your results are irrefutable, using the "difference blend mode litmus test." I hope I didn't give anyone the impression that I think otherwise.
My use of hyperbole was to draw attention to information which might not have been consumed otherwise. If that bothered you, I apologize.
ezpkns wrote: Why do I use High Pass over USM? Are you testing me, or actually curious? It's better, that's why. Good enough?
If you're implying that if I'm looking for a way to sharpen high-frequency information separately, you have a better way, I think I acknowledged that in my original post.
I'm curious because USM gives the same result as the 'default' sharpening discussed in the HP thread, and genuinely wonder how the HP filter works better into your workflow? Obviously we all do things differently, but I'm always curious about the way others go about things.
Fernando L Pacheco wrote: That's just the thing. JPG is 8bits per pixel (0-255) or 256 discrete values
A 12bit RAW file is 4096 discrete values
A 14bit RAW file is 16,384 discrete values
A 16bit RAW file is 65,536 discrete values
HUGE difference
If you divide in you'll see that a 14bit RAW has 64 TIMES the number of possible tonal values per pixel than an 8bit jpeg. Nevermind the compression artifacting and other issues.
btw, how do you get Photoshop to show you those values, I have looked everywhere and that damn info tab only shows 8bit values
Its great for landscapes and buildings, stuff with detail i think its a quick and lazy way to sharpen but it works well... I use it sometimes on models and layer/paint/opacity...till you get what you like..
i use a HP sharpen about 50% of the time for digital work on the web. Usually on a new layer, made with the channel that i find has the best attributes for what i want to do (i.e. eyes, skin, etc etc). Painted in and selectively used.
I do use it 100% of the time just before i output on the inkjet. It adds a nice level of sharpness and really makes the image pop from the page when it comes out of my 4000. I don't use it so much when i have photographic prints made.
The only thing that bothers me is that we seem to have gotten off on the wrong foot. I think you read something I wrote as being facetious when it wasn't, or something. I have the utmost respect for you and all you've done to share your discoveries. I was very busy with a long and dreary job when you first started that thread, but I followed along as best I could, but couldn't afford the time to experiment.
I'm fascinated by your work, as little as I understand it - it's like Deke McTalkstoomuch's sharpening video goes to the 5th dimension. I just couldn't spare the time to be in the thick of things when the initial experimentation was being done, and now I'm still a little behind.
However, as you can tell from the responses so far, lots of people are still using high pass - outside of MM, the percentage is probably much higher - and the version I'm most familiar with is stamping visible layers to a new layer, and running high pass on that to taste, using one of the overlay blend modes to sharpen the underlying layers. Since it behooves one to desaturate the HP layer anyway, I started experimenting with using different channels in lieu of a desaturated merged copy.
I eventually settled on the red channel most of the time, because of it's lack of contrast in the skin tones, which makes it ideal for sharpening everything but pores, etc. I don't understand your technique well enough to adapt this method for use without the High Pass filter, but it works well enough I thought I'd share it anyway, since I've never heard anyone talk about it.
ezpkns wrote: I eventually settled on the red channel most of the time, because of it's lack of contrast in the skin tones, which makes it ideal for sharpening everything but pores, etc. I don't understand your technique well enough to adapt this method for use without the High Pass filter, but it works well enough I thought I'd share it anyway, since I've never heard anyone talk about it.
That marks three failures on my part, then. To use the red channel in the manner you describe, you would simply make a merged copy of it, paste it back in as a layer, and treat it just like you would any other layer for sharpening - the HP thread technique would extract the information just like the HP filter does, and with the same result.
And besides, High Pass really doesn't suck.
As a method, no, certainly not. But the filter's busted.
Nice one. I do scripts, too. Maybe a new name for me? Scraption Man? Scription Man? Scraction Man?
I personally don't like the way my Camera's handle color rendition or sharpening when processing JPG's. It's not to my taste.
