Yes but it's pretty darn good considering the extremes of the image :rutt
How did I get that old man with the cane and beard named after me? Andy, how old are you? And cranky?
Anyway, I have a sort of long time thing I've been trying to invent. 1/2 algorithm and 1/2 person. Sort of a Cyborg photoshop and/or i2e thing. It uses human judgment to control very sophisticated post processing algorithms in a series of QA steps.
Is this sky? Skin? Water?
Want the glamor look?
How old is this person?
Is there something neutral?
Is the the lightest point of interest?
Too dark?
Maybe 1 minute not 5 seconds and definitely not hours. Very deterministic. Best results possible. Now how much would you pay?
Great. I'm sick of that darn image of yours anyway. I'm hoping somebody like Edgework might take a swing at it. Or perhaps somebody has gotten good at lightzones? This might be a good image for that approach.
Great. I'm sick of that darn image of yours anyway. I'm hoping somebody like Edgework might take a swing at it. Or perhaps somebody has gotten good at lightzones? This might be a good image for that approach.
Glad you liked it...
Apart from the last smiley-related OT this has been a great thread so far. for me at least. I learned several new things, which does not happen often
I agree with you that part of this thread has to do with Chapter 2 in particular, and there should probably be a link from there to here. But there are other parts of the image treatment that have to do with false profile, luminosity blending, steepening the K, and color evaluation. So while I agree that the "content" parts of this thread deserve a nod from Chapter 2 of PP5, I think this thread could also stand archiving for reference in the discussion of other chapters. Much of the banter could be stripped without any loss for purposes of PP5, as could these discussions of where the thread belongs.
I agree with you that part of this thread has to do with Chapter 2 in particular, and there should probably be a link from there to here. But there are other parts of the image treatment that have to do with false profile, luminosity blending, steepening the K, and color evaluation. So while I agree that the "content" parts of this thread deserve a nod from Chapter 2 of PP5, I think this thread could also stand archiving for reference in the discussion of other chapters. Much of the banter could be stripped without any loss for purposes of PP5, as could these discussions of where the thread belongs.
I took one more swing at this. You might want to download and look in photoshop because I cut it very close between showing the detail in the fabric and keeping the uniforms dark. Looks much better in PS than in Firefox for some reason I don't really understand.
I took a much more conservative approch than before.
Shadow/highlights on a layer to bring out fabric detail. Very low tonal width (about 0). Turned town opacity until the details were only just visible.
Used Dan's new inverted/blurred/overlay technique from PP5E, Ch 18. Used lighten and darken layers from the original to turn down the lighten and darkening effect separately.
To CMYK and steepen the K curve radically to bring out the folds in the uniforms and reestablish that dark blue.
Keep CMYK duplicate and move to LAB.
Fix green cast in the sky and pants with a command-click curves layer.
Fix the magenta cast in the faces and enhance all the colors with a second command-click layer. Try to get the L curve steep through the dark-midtones for the faces. Use the K channel from the CMYK duplicate as a layer mask to keep the shadows deep and more neutral.
Sharpen USM and HIRALOAM on separate layers and blend with sliders and blend-if sliders.
It looks good but is still too dark for my taste. Here are my other quibbles:
The officer's face is still more magenta than yellow. The rest of the troop have about equal magenta and yellow.
There's now some green in the apron shadows.
The fill in the near blown sky just over their heads on the right is overdone for my taste. It looks more like a picture of the sky now than of the troops. (My problem with your first correction is that on first blush, it looked like a picture of the flag, with the troops as an afterthought. This one, because of the tonal balance, is still to focused away from where we should be looking.)
Here's something I've been tussling with in this photo. In the original, the sky is definitely too green. And the faces are definitely too magenta. The use of fill flash might explain the conflicting casts. But what about the trousers? They are also too green (everyone seems to think), but they presumably would get some of the same magenta cast that the faces got if the flash was the cause. So one of the following must be true: 1) the faces really are that magenta, perhaps from overheating on a march, 2) The trousers are actually bluish green, or 3) there is some other factor that's causing strange casting that I don't see or understand. Or perhaps there is a combination of the three.
I see a difference between the firefox version and the PS version. And I too cannot explain it.
I took one more swing at this. You might want to download and look in photoshop because I cut it very close between showing the detail in the fabric and keeping the uniforms dark. Looks much better in PS than in Firefox for some reason I don't really understand.
I took a much more conservative approch than before.
Shadow/highlights on a layer to bring out fabric detail. Very low tonal width (about 0). Turned town opacity until the details were only just visible.
Used Dan's new inverted/blurred/overlay technique from PP5E, Ch 18. Used lighten and darken layers from the original to turn down the lighten and darkening effect separately.
To CMYK and steepen the K curve radically to bring out the folds in the uniforms and reestablish that dark blue.
Keep CMYK duplicate and move to LAB.
Fix green cast in the sky and pants with a command-click curves layer.
Fix the magenta cast in the faces and enhance all the colors with a second command-click layer. Try to get the L curve steep through the dark-midtones for the faces. Use the K channel from the CMYK duplicate as a layer mask to keep the shadows deep and more neutral.
Sharpen USM and HIRALOAM on separate layers and blend with sliders and blend-if sliders.
My favorite so far. What do you think?
For whatever reason I noticed that I cannot stand the slightest hint of Shadows/Highlights It has certain tell-tell signs that immediately kills the whole thing for me.
I wonder how this image would look like if you omit that dreaded step?
Oh, and I like Andy's version of the sky much better (the one that i2e made from raw).
So one of the following must be true: 1) the faces really are that magenta, perhaps from overheating on a march, 2) The trousers are actually bluish green, or 3) there is some other factor that's causing strange casting that I don't see or understand. Or perhaps there is a combination of the three.
Duffy
I'd say it was mainly #1. It was pretty darn hot that days (November is a still a summer month here in SoCal) and the poor souls were standing under the scorching sun for at least an hour.
Don't you just like this picture? Two totally different light sources, and every single important color is called for..
Assuming that you are right, and they are flushed from the heat, the next question is what to do in a correction. To a certain extent, it would depend on the purpose of the shot. If you are trying to show the heat of the day, and the effort, then by all means make them flushed and a bit sunburned. But if you are trying to sell the participants these shots, then you would probably want to eliminate some of the magenta, even if it was there. And you might take a third approach if you were trying to pretend that these guys were actual civil war troops (they would have been more weatherbeaten, and probably tanned). So that's an area of choice and taste, to a certain extent.
I've also been thinking about Rutt's preoccupation with the uniform details. If this was a picture of an actual group of soldiers marching, the focus would be much more on the faces, their expressions, the people themselves. But its a re-enactment, and so I think the uniforms probably matter almost as much, if not as much as the people. But, if this was a picture of, say, troops in Afghanistan (sp?), I don't think we would be as likely to be concerned with the folds in their uniforms, because the power of the shot would much more likely be in their faces and postures. Tio a certain extent, we could take the uniforms for granted. In a re-enactment, the uniforms are, however, very much the point of the exercise as much as anything else.
And yes, I think its a very good shot for studying. That's why I've spent so much time thinking about it.
Can't help your shadow/highlights heebeegeebees, Nik, but really I don't think that's what you are seeing. I limited it to just those folds. More likely you didn't like the radius of the inverted overlay blend or even the flatness of the raw conversion I started with (flat curves, 0 contrast). You cold try another approach and do 3 conversions and treat like HDR and see what you get. I've been on record saying that's not the greatest idea, but in this case it might be. I also sent this image to Fabio, the father of Lightzones. Maybe he'll see what he can do.
Ah, two light sources! Fill flash, but not enough. Probably there isn't enough fill flash in the world for that So Cal sky.
OK, you guys are right about the sky. And I was stupid. I omitted a Dan standard: red channel L blend of skies. It really never fails to be a good thing:
I just did this as a final step. I suppose it's an example of fine tuning at the very end after sharpening. Ooops, might have been better to do earlier, but better to do last than not at all. Same with a little B channel burn for the faces and sponge for the apron. Very subtle, but looks a little healthier.
As Nik points out, part of what is so strange about this shot is that these guys aren't dressed for the weather. I've found that it's ok to have some magenta > yellow in the flesh as long there is a lot of flesh where it's the other way around. Exercise, hot or cold weather, homeless people, that's where you find this most. Could fix that flesh in the officer with a
I played with your last version a little bit. Basically, I put a 1.4 gamma version on a new layer, in luminosity mode, at 40% opacity, and with an inverted L channel mask. It looked better in PS. The troops and uniforms were lightened up some without messing with the sky or flag.
Then I posted it, and it looked worse. Something strange is going on between PS and Firefox. These were side by side images. In PS, the change was a definite improvement to my eye, but in the browser it was worse. So I deleted it. Go figure.
Then I posted it, and it looked worse. Something strange is going on between PS and Firefox. These were side by side images. In PS, the change was a definite improvement to my eye, but in the browser it was worse. So I deleted it. Go figure.
Duffy
Looks especially bad in linux firefox. I wish someone would explain this ff vs ps issue. David? Baldy?
Looks especially bad in linux firefox. I wish someone would explain this ff vs ps issue. David? Baldy?
FF is not color space aware. Make sure you're converting to sRGB before posting/uploading. If you're saving in aRGB or some other color space, nothing will look right. Baldy has some great posts on this, especially relating to Safari (goes to find it....)
I wish this were the problem, but I don't think it is. This image was uploaded in sRGB and looks different in FF and PS and even worse in linux FF.
This happens to me once in a blue moon and nobody has ever really explained it.
One theory that I've heard is that PS is not only aware of the profile of the image but also of the monitor. I don't really understand this, but perhaps it means something.
FF is not color space aware. Make sure you're converting to sRGB before posting/uploading. If you're saving in aRGB or some other color space, nothing will look right. Baldy has some great posts on this, especially relating to Safari (goes to find it....)
Can't help your shadow/highlights heebeegeebees, Nik, but really I don't think that's what you are seeing. I limited it to just those folds. More likely you didn't like the radius of the inverted overlay blend or even the flatness of the raw conversion I started with (flat curves, 0 contrast). You cold try another approach and do 3 conversions and treat like HDR and see what you get. I've been on record saying that's not the greatest idea, but in this case it might be. I also sent this image to Fabio, the father of Lightzones. Maybe he'll see what he can do.
Ah, two light sources! Fill flash, but not enough. Probably there isn't enough fill flash in the world for that So Cal sky.
OK, you guys are right about the sky. And I was stupid. I omitted a Dan standard: red channel L blend of skies. It really never fails to be a good thing:
I just did this as a final step. I suppose it's an example of fine tuning at the very end after sharpening. Ooops, might have been better to do earlier, but better to do last than not at all. Same with a little B channel burn for the faces and sponge for the apron. Very subtle, but looks a little healthier.
As Nik points out, part of what is so strange about this shot is that these guys aren't dressed for the weather. I've found that it's ok to have some magenta > yellow in the flesh as long there is a lot of flesh where it's the other way around. Exercise, hot or cold weather, homeless people, that's where you find this most. Could fix that flesh in the officer with a
Sky is much better now, but the rest... Maybe it's the profile, IE7 and some other things all mixed together, but all I see now are the halos and other signs of a major sharpening done, most likely hiraloam in biblical proportions...
This treatment may have followed all the designated steps, but as the end result I'd say I like my original more (except for the skies)
Sky is much better now, but the rest... Maybe it's the profile, IE7 and some other things all mixed together, but all I see now are the halos and other signs of a major sharpening done, most likely hiraloam in biblical proportions...
This treatment may have followed all the designated steps, but as the end result I'd say I like my original more (except for the skies)
Upon looking on a good monitor (and in CS3, not IE), I can honestly say: your treatment is great!
It better be. Andy spent 2x5 seconds on his two versions. I spent the better part of the weekend and then thought about it for during my commute and on the elliptical trainer and then redid while my wife watched The Antiques Raodshow. I'd say I lost the resuls / effort contest big time. But unlike I2E, I did actually learn something, so I guess that's something.
Now the real question is: would you be willing to provide the whole exact detailed series of steps along with the the layer palette/dialogs captures?
You had to ask that, didn't you? I am soooo sick of processing this particular image. But I have two good ideas:
Let's see if I can lead YOU through the steps without giving you the exact value of every slider and opacity setting. Then you can post it if you like. I think we can get pretty darn close and probably you'll like it even better when we are done. I can help a little. Here is my xmp file for it, so you can start with my flat raw conversion.
Why not contribute this image to Dan to use in his classes? It's perfect torture for that purpose. I'll bet you'll end up with some great treatments over time? Email me if you are interested.
(Funny how the hardware and software makes the difference, though:-)
We understand a lot of this, but not all. Looking at it at an even multiple of it's actual size we already know is important for evaluating sharpening. Good monitors display shades where bad monitors show banding. Combine these and aggressive sharpening is going to look way overdone.
What we don't understand is why this particular image looks so much darker in FF (and I assume IE) than in PS. My current theory is that I took the shadows so close to the edge that PS's knowledge of the monitor profile is relevant.
Nik, do you have a professional ink jet printer? Which RIP? If so, try printing and see how it looks. If not, email me and I'll print for you with ImagePrint and my Epson 4k. My guess is that this version will look pretty good, but it would be interesting to see how it compares to the I2E version when printed.
Comments
How did I get that old man with the cane and beard named after me? Andy, how old are you? And cranky?
Anyway, I have a sort of long time thing I've been trying to invent. 1/2 algorithm and 1/2 person. Sort of a Cyborg photoshop and/or i2e thing. It uses human judgment to control very sophisticated post processing algorithms in a series of QA steps.
Maybe 1 minute not 5 seconds and definitely not hours. Very deterministic. Best results possible. Now how much would you pay?
Catapultam habeo. Nisi pecuniam omnem mihi dabis, ad caput tuum saxum immane mittam
http://www.mcneel.com/users/jb/foghorn/ill_shut_up.au
Great. I'm sick of that darn image of yours anyway. I'm hoping somebody like Edgework might take a swing at it. Or perhaps somebody has gotten good at lightzones? This might be a good image for that approach.
Glad you liked it...
Apart from the last smiley-related OT this has been a great thread so far. for me at least. I learned several new things, which does not happen often
I agree with you that part of this thread has to do with Chapter 2 in particular, and there should probably be a link from there to here. But there are other parts of the image treatment that have to do with false profile, luminosity blending, steepening the K, and color evaluation. So while I agree that the "content" parts of this thread deserve a nod from Chapter 2 of PP5, I think this thread could also stand archiving for reference in the discussion of other chapters. Much of the banter could be stripped without any loss for purposes of PP5, as could these discussions of where the thread belongs.
Duffy
I took one more swing at this. You might want to download and look in photoshop because I cut it very close between showing the detail in the fabric and keeping the uniforms dark. Looks much better in PS than in Firefox for some reason I don't really understand.
I took a much more conservative approch than before.
My favorite so far. What do you think?
The officer's face is still more magenta than yellow. The rest of the troop have about equal magenta and yellow.
There's now some green in the apron shadows.
The fill in the near blown sky just over their heads on the right is overdone for my taste. It looks more like a picture of the sky now than of the troops. (My problem with your first correction is that on first blush, it looked like a picture of the flag, with the troops as an afterthought. This one, because of the tonal balance, is still to focused away from where we should be looking.)
Here's something I've been tussling with in this photo. In the original, the sky is definitely too green. And the faces are definitely too magenta. The use of fill flash might explain the conflicting casts. But what about the trousers? They are also too green (everyone seems to think), but they presumably would get some of the same magenta cast that the faces got if the flash was the cause. So one of the following must be true: 1) the faces really are that magenta, perhaps from overheating on a march, 2) The trousers are actually bluish green, or 3) there is some other factor that's causing strange casting that I don't see or understand. Or perhaps there is a combination of the three.
I see a difference between the firefox version and the PS version. And I too cannot explain it.
Duffy
For whatever reason I noticed that I cannot stand the slightest hint of Shadows/Highlights
I wonder how this image would look like if you omit that dreaded step?
Oh, and I like Andy's version of the sky much better (the one that i2e made from raw).
I'd say it was mainly #1. It was pretty darn hot that days (November is a still a summer month here in SoCal) and the poor souls were standing under the scorching sun for at least an hour.
Don't you just like this picture?
I've also been thinking about Rutt's preoccupation with the uniform details. If this was a picture of an actual group of soldiers marching, the focus would be much more on the faces, their expressions, the people themselves. But its a re-enactment, and so I think the uniforms probably matter almost as much, if not as much as the people. But, if this was a picture of, say, troops in Afghanistan (sp?), I don't think we would be as likely to be concerned with the folds in their uniforms, because the power of the shot would much more likely be in their faces and postures. Tio a certain extent, we could take the uniforms for granted. In a re-enactment, the uniforms are, however, very much the point of the exercise as much as anything else.
And yes, I think its a very good shot for studying. That's why I've spent so much time thinking about it.
Duffy
Ah, two light sources! Fill flash, but not enough. Probably there isn't enough fill flash in the world for that So Cal sky.
OK, you guys are right about the sky. And I was stupid. I omitted a Dan standard: red channel L blend of skies. It really never fails to be a good thing:
I just did this as a final step. I suppose it's an example of fine tuning at the very end after sharpening. Ooops, might have been better to do earlier, but better to do last than not at all. Same with a little B channel burn for the faces and sponge for the apron. Very subtle, but looks a little healthier.
As Nik points out, part of what is so strange about this shot is that these guys aren't dressed for the weather. I've found that it's ok to have some magenta > yellow in the flesh as long there is a lot of flesh where it's the other way around. Exercise, hot or cold weather, homeless people, that's where you find this most. Could fix that flesh in the officer with a
I played with your last version a little bit. Basically, I put a 1.4 gamma version on a new layer, in luminosity mode, at 40% opacity, and with an inverted L channel mask. It looked better in PS. The troops and uniforms were lightened up some without messing with the sky or flag.
Then I posted it, and it looked worse. Something strange is going on between PS and Firefox. These were side by side images. In PS, the change was a definite improvement to my eye, but in the browser it was worse. So I deleted it. Go figure.
Duffy
Looks especially bad in linux firefox. I wish someone would explain this ff vs ps issue. David? Baldy?
FF is not color space aware. Make sure you're converting to sRGB before posting/uploading. If you're saving in aRGB or some other color space, nothing will look right. Baldy has some great posts on this, especially relating to Safari (goes to find it....)
.....and here it is:
http://www.dgrin.com/showthread.php?t=52712
Dgrin FAQ | Me | Workshops
This happens to me once in a blue moon and nobody has ever really explained it.
One theory that I've heard is that PS is not only aware of the profile of the image but also of the monitor. I don't really understand this, but perhaps it means something.
Sky is much better now, but the rest... Maybe it's the profile, IE7 and some other things all mixed together, but all I see now are the halos and other signs of a major sharpening done, most likely hiraloam in biblical proportions...
This treatment may have followed all the designated steps, but as the end result I'd say I like my original more (except for the skies)
Download orignal and look in PS. How different?
Will do tonight.
When you do, also make sure to look at an even size multiplication (25%, 50%)
Mea culpa, mea culpa
Upon looking on a good monitor (and in CS3, not IE), I can honestly say: your treatment is great!
(Funny how the hardware and software makes the difference, though:-)
Now the real question is: would you be willing to provide the whole exact detailed series of steps along with the the layer palette/dialogs captures?
It better be. Andy spent 2x5 seconds on his two versions. I spent the better part of the weekend and then thought about it for during my commute and on the elliptical trainer and then redid while my wife watched The Antiques Raodshow. I'd say I lost the resuls / effort contest big time. But unlike I2E, I did actually learn something, so I guess that's something.
You had to ask that, didn't you? I am soooo sick of processing this particular image. But I have two good ideas:
We understand a lot of this, but not all. Looking at it at an even multiple of it's actual size we already know is important for evaluating sharpening. Good monitors display shades where bad monitors show banding. Combine these and aggressive sharpening is going to look way overdone.
What we don't understand is why this particular image looks so much darker in FF (and I assume IE) than in PS. My current theory is that I took the shadows so close to the edge that PS's knowledge of the monitor profile is relevant.
Nik, do you have a professional ink jet printer? Which RIP? If so, try printing and see how it looks. If not, email me and I'll print for you with ImagePrint and my Epson 4k. My guess is that this version will look pretty good, but it would be interesting to see how it compares to the I2E version when printed.