Using ACR after scan stage

[Mark D Segal]

Linear data doesn’t just look pretty flat; it looks extremely dark with horrible color.  That’s why with camera raw files LR/ACR provides a gamma adjusted view for editing photos.  I don’t think it’s practical to make color and tone adjustments directly on and viewing linear data.  But if Jeff and you are able to do so, I must be wrong.  

By the way, what profile would be embedded with a linear file?

Dean, just to make sure I'm not talking through my hat - one always likes to question oneself despite years of experience with this, because software and possibilities keep changing. I just now opened my trusty Nikon 5000-ED, hauled out a test slide (just an ordinary photo that doesn't have extremes of contrast) and ran four scans: gamma 1.0 profiled and unprofiled; gamma 2.2 profiled and unprofiled. SilverFast allows one to make all those choices. Indeed the gamma 1.0 scans look like you say they would look - as usual. Profiling or not profiling in this case didn't make a big difference to the starting point whether in gamma 1.0 or gamma 2.2. I opened them all in Lightroom 4.3 and I was able to quickly bring all of them to roughly the same appearance with different treatment in each case. What works better in the final analysis can also be image-dependent. This is why my essential point remains that one needs to remain pragmatic about all this and recognize each software for its comparative advantage.

[dmerger]

Jeff, where you said above "very open flat scans" I thought you had in mind linear because that's what you get from a gamma setting of 1.0 in the scanner; however one can also produce open flat scans that are not necessarily linear so your point is well taken.

Dean's point that linear scans start life looking dark and ugly is correct, but in the newer versions of LR/ACR they can be handled well even though when you first look at them it's depressing.

Mark, you must live in a world different than mine.    Linear scans are not open and flat looking, not even close. Attached is a sample of a linear scan and the same scan, but with my linear profile assigned.  (Both versions were converted to sRGB for web viewing and saved as jpegs.) 

It’s incredible that you or anyone else would find it practical to use that linear version to make color and tone adjustments without first making a gamma adjustment.  It seems to defy belief, but if you say you do so, I’ll take your word on it. 

[Mark D Segal]

We were probably writing on the same subject at very close intervals. Yes I agree, those linear scans look like crap before you've done anything to them, but the point is that they contain all the image data you need for post-scan adjustment into decent images. As I said, not usually my favorite starting point either because they're very dark and need radical remapping, but they can be worked into perfectly satisfactory images, depending on the image and the software. My normal workflow is to make gamma-adjusted scans as well, so in practice we aren't far apart.

[digitaldog]

Linear captures do not necessarily look nor have to appear dark. They do when you view them assumed to be gamma corrected. If you have a profile that properly defines the linear gamma data, it looks quite good. I have really old Kodak DCS captures in a linear form but after assigning the proper ICC profile, it looks just fine. I've used that example for years when showing people that what first appears dark and awful can appear lovely without changing a single numeric value in the document.

[roa5100xx]

Hi Everyone,
Thank you for all of your replies, this turned out to be a very interesting discussion. I see I have a lot to learn.

I have been using SilverFast for about two(maybe three ) weeks. I have been using ARC and Photoshop for long time, and feel more comfortable using it. That's why I asked the question. But I don't want to give up anything (image quaity) that I should be doing in SilverFast by using ARC. Like I said "I have a lot to learn." Well, I guess there are no easy answers.

It seems I have a number options. So I will do some experimenting and see which options I like best.

Thanks again

Herbert
 

[roa5100xx]

Mark,
I download your book, very good job.

Herbert

[Mark D Segal]

Mark,
I download your book, very good job.

Herbert

Much appreciated Herbert, glad you find it useful.

[Mark D Segal]

Linear captures do not necessarily look nor have to appear dark. They do when you view them assumed to be gamma corrected. If you have a profile that properly defines the linear gamma data, it looks quite good. I have really old Kodak DCS captures in a linear form but after assigning the proper ICC profile, it looks just fine. I've used that example for years when showing people that what first appears dark and awful can appear lovely without changing a single numeric value in the document.

Andrew, while I appreciate what you are saying here in its context, the fact is that for people using SilverFast 8 and its profiles, there are numerous ways of doing this, between using the HDR scan mode or not using it, profiling or not profiling, gamma linear or gamma>1.0, and if using HDR scan mode, gamma >1.0 selected or not selected for HDR output to an external editor. Within either the HDR or non-HDR scan modes, the choices are independent variables. There are a total of 12 permutations and combinations I can develop within SilverFast before we get to an external editor. Depending on what you select, you will start-out with a dark image or a nice looking image assuming the original media is nice looking. Depending on the photograph itself, I can make edits to produce a similar-looking and decent end-result from any of these options be it within SilverFast alone, or SilverFast combined with LR, or LR alone, depending on the challenges of the image.  I'm not about to start writing and illustrating another dissertation on this because I just don't have time; but what I'm saying here is why I recommend that we should be pragmatic. There are many answers and we can certainly provide guidance on what we think are best practices, but it is up to users to then know what the options are and to experiment and develop their own workflow preferences.

[dmerger]

Linear captures do not necessarily look nor have to appear dark. They do when you view them assumed to be gamma corrected. If you have a profile that properly defines the linear gamma data, it looks quite good. I have really old Kodak DCS captures in a linear form but after assigning the proper ICC profile, it looks just fine. I've used that example for years when showing people that what first appears dark and awful can appear lovely without changing a single numeric value in the document.

Andrew, maybe I’m just missing the point.  Your example appears to show the same thing as my example.  Your first example is dark.  In your second example, you assigned your profile and it looks normal.  I did the same.  Is there something about your example that is different than mine?

Also, I don’t understand what you mean by “without changing a single numeric value in the document.”  When I look at your example in PS, the numeric values in the two are very different.  

[digitaldog]

Also, I don’t understand what you mean by “without changing a single numeric value in the document.”  When I look at your example in PS, the numeric values in the two are very different.  

The only thing I did was assign the correct profile to the image. That doesn't change the numbers at all. Only the meaning of the numbers.

[Schewe]

Where you say you ignore profiles with the Epson scanner, I find if not using a scanner profile at the scan stage, the image starts life more corrected in the post-scan stage of image editing by assigning the scanner profile on opening it in Photoshop and converting to the working space.

I'm not "ignoring" profiles with the Epson, I've just never bothered to do a custom scanner profile. I have the Epson scanner software simply scan into Adobe RGB. For the Imacon, I've made a really good custom profile using Don Huchison's Q60 target and ProfileMaker (I've not bothered to redo it in i1 Profiler since I rarely need to scan).

[Alan Klein]

1. I use a V600 with the supplied Epson scan software.   You can't do multiple scans or hardware adjustment of the scan itself like a V750. I find that if I set Color Correct-Continuous Auto Exposure with a Gamma of 2.2 (default setting), I get color restoration in the scan that is very good.  Are the other Gamma settings of any value?  Other is 1.8  You can also select ICM instead with either sRGB or Adobe RGB, or Colormatch, or Apple RGB outputs.  I've never tried it but was wondering the value of using ICM instead of the aforementioned Continuous Auto Exposure?  

2.  How do you adjust LR3  when I scan flat. (I usually get a dark off color scan that's in a narrow band of the histogram).   If I use Auto Levels adjustment in Elements 8 on a flat scan, I can match Auto control during the scan mentioned in 1. above.  However, I'm concerned that Elements 8 may be clipping with Auto levels.  How would I use Elements to avoid clipping?

3.  I also have LR3.  How would I use LR3 on a flat scan?

[Mark D Segal]

I'm not "ignoring" profiles with the Epson, I've just never bothered to do a custom scanner profile. I have the Epson scanner software simply scan into Adobe RGB. For the Imacon, I've made a really good custom profile using Don Huchison's Q60 target and ProfileMaker (I've not bothered to redo it in i1 Profiler since I rarely need to scan).

OK, where you said above "(and ignore "profiles")" I took it literally. I've been relying on LSI's profiles or their Auto IT-8 custom profiling using their targets, which are IT8's, less granular than Don's. It would be nice to compare - one of these days when I get some more recent profiling software.

[Mark D Segal]

1. I use a V600 with the supplied Epson scan software.   You can't do multiple scans or hardware adjustment of the scan itself like a V750. I find that if I set Color Correct-Continuous Auto Exposure with a Gamma of 2.2 (default setting), I get color restoration in the scan that is very good.  Are the other Gamma settings of any value?  Other is 1.8  You can also select ICM instead with either sRGB or Adobe RGB, or Colormatch, or Apple RGB outputs.  I've never tried it but was wondering the value of using ICM instead of the aforementioned Continuous Auto Exposure?  

2.  How do you adjust LR3  when I scan flat. (I usually get a dark off color scan that's in a narrow band of the histogram).   If I use Auto Levels adjustment in Elements 8 on a flat scan, I can match Auto control during the scan mentioned in 1. above.  However, I'm concerned that Elements 8 may be clipping with Auto levels.  How would I use Elements to avoid clipping?

3.  I also have LR3.  How would I use LR3 on a flat scan?

I'm a control freak so I don't do auto-anything in any software. Any time of I've tried, I've never been happy with the results, either because of clipping or some other issue, so I need to undo it and then adjust the photo properly. So I just stopped even thinking about auto this or that and adjust them manually. In LR3 you would use the Exposure, Fill Light sliders, increasing both, and possibly reducing contrast. You should consider upgrading to LR4.3 - as good as LR 3 was, 4.3. is that much better for this kind of work - it will give you even more granular and more flexible control over darks and lights. and the upgrade price is reasonable.

[dmerger]

Profiling or not profiling in this case didn't make a big difference to the starting point whether in gamma 1.0 or gamma 2.2.

Mark, if assigning a profile to your linear scan didn’t make a dramatic difference, then you must have used the wrong profile or a very poor profile.  See Andrew’s and my examples above.  A proper linear profile should make your image look normal.

[dmerger]

Unless the scanner has user-adjustable lamp-control, exposure adjustment is not a scanner dependent operation. It is software controlled.

Mark, I didn’t reply this statement before because, although it’s wrong, it seemed to be irrelevant, and I didn’t see any reason to debate a point neither I nor Herbert raised.  But on reflection, maybe there is some use in discussing it.

My scanner and scanning software (Minolta 5400) has an exposure adjustment.  It works similar to camera exposure adjustment.  It controls the scan time.  The physical effect is very obvious.  With increases or decreases in exposure the scan time can vary significantly. 

I know that you also own a Minolta 5400.  So I wondered why you’d make the false assertion about exposure adjustments.  How could anyone who used that scanner not notice the very prominent control and the very obvious change in scan times? 

Then it occurred to me that you mainly use SilverFast.  Perhaps SilverFast doesn’t implement the Minolta 5400’s exposure adjustment, so you weren’t aware that it existed? 

I’ve also used SilverFast with my scanner, but it has been a while.  I recall that SilverFast could not access my scanner’s manual focus controls, but I’m not sure about the hardware exposure control.

I found it a little frustrating that such expensive software that was a version specifically designed to operate my scanner couldn’t access the manual focus control. I could have put up with that flaw, however, had I otherwise liked the results I got with SilverFasts, but I didn’t.  As a result, I haven’t bothered to install it on my current system, so I can’t check now to see if it also failed to implement the hardware exposure control.  If so, however, it would explain why you weren’t aware that some scanners (including a scanner you own) have hardware exposure controls that control the scan time.

In any event, my point is that if a scanner has an actual hardware exposure control, it would be better to optimize the hardware exposure rather than just rely on software exposure adjustments.  The reason is the same reason that you’d want to optimize exposure in a camera rather than just rely on post processing.

[Mark D Segal]

Mark, if assigning a profile to your linear scan didn’t make a dramatic difference, then you must have used the wrong profile or a very poor profile.  See Andrew’s and my examples above.  A proper linear profile should make your image look normal.

Neither, but there is a finer point in this I need to investigate and I shall revert on it when ready.

[Fine_Art]

Mark, you must live in a world different than mine.    Linear scans are not open and flat looking, not even close. Attached is a sample of a linear scan and the same scan, but with my linear profile assigned.  (Both versions were converted to sRGB for web viewing and saved as jpegs.) 

It’s incredible that you or anyone else would find it practical to use that linear version to make color and tone adjustments without first making a gamma adjustment.  It seems to defy belief, but if you say you do so, I’ll take your word on it. 

If I apply a gamma 2.2 to your bright red image I do not get your more normal looking photo. Vise versa. You must be changing something else.

[Mark D Segal]

Mark, I didn’t reply this statement before because, although it’s wrong, it seemed to be irrelevant, and I didn’t see any reason to debate a point neither I nor Herbert raised.  But on reflection, maybe there is some use in discussing it.

My scanner and scanning software (Minolta 5400) has an exposure adjustment.  It works similar to camera exposure adjustment.  It controls the scan time.  The physical effect is very obvious.  With increases or decreases in exposure the scan time can vary significantly. 

I know that you also own a Minolta 5400.  So I wondered why you’d make the false assertion about exposure adjustments.  How could anyone who used that scanner not notice the very prominent control and the very obvious change in scan times? 

Then it occurred to me that you mainly use SilverFast.  Perhaps SilverFast doesn’t implement the Minolta 5400’s exposure adjustment, so you weren’t aware that it existed? 

I’ve also used SilverFast with my scanner, but it has been a while.  I recall that SilverFast could not access my scanner’s manual focus controls, but I’m not sure about the hardware exposure control.

I found it a little frustrating that such expensive software that was a version specifically designed to operate my scanner couldn’t access the manual focus control. I could have put up with that flaw, however, had I otherwise liked the results I got with SilverFasts, but I didn’t.  As a result, I haven’t bothered to install it on my current system, so I can’t check now to see if it also failed to implement the hardware exposure control.  If so, however, it would explain why you weren’t aware that some scanners (including a scanner you own) have hardware exposure controls that control the scan time.

In any event, my point is that if a scanner has an actual hardware exposure control, it would be better to optimize the hardware exposure rather than just rely on software exposure adjustments.  The reason is the same reason that you’d want to optimize exposure in a camera rather than just rely on post processing.

No, the statement isn't wrong and I'm well aware that some scanners provide for hardware-based exposure control and others don't. The fact is that unless the scanner provides for exposure control (a.k.a. "lamp control") in the hardware, all exposure adjustments are software-based - i.e. remapping pixels. Lamp control in the hardware can be managed in two ways depending on how the scanner was designed: either the media passes over the lamp or the lamp over the media slower/faster for increasing/decreasing exposure, or lamp brightness is actually changed, or a combination of two. More generally it would be speed controlled. This is a matter I have discussed with engineers who work with the designs of many of these scanners for purposes of developing control algorithms in the scanning application.

I know what the Minolta 5400 does because I own one, but I wasn't referencing any particular scanner in my comment. It was a general point in relation to something you said above about exposure control; maybe I misunderstood you. Anyhow, turning to the Minolta 5400, it's now history at least as far as SilverFast is concerned. As you know, Minolta vacated this market segment some years ago and sold it (to Sony I think). There have been no driver upgrades from Minolta or Sony. Whereas LaserSoft Imaging developed a totally up-dated driver for the Nikon scanners (also no longer being made, but that is more recent), they have not done so for the Minolta scanner; in the SilverFast world you can use this Minolta scanner with version 6.6. up to Windows 2000 or Mac OSX 10.3; Vuescan supports it on current operating systems. While the Minolta 5400 produces excellent results, it is very slow. The results are at least as good from the Nikon 5000 and it is a much more efficient device. I am now on Mac OSX 10.6.8 and SilverFast 8, so the Minolta scanner is packed away. The two scanner's I'm using are the Nikon Super Coolscan 5000-ED for 35mm, and the Epson V-750 Pro for other stuff. The Nikon scanner does have lamp control and SilverFast 8 provides for it. The Nikon scanner does also have focusing control, and SilverFast 8 provides extensive control over it either automatically or manually, including the selection of focal points.

[dmerger]

If I apply a gamma 2.2 to your bright red image I do not get your more normal looking photo. Vise versa. You must be changing something else.

All I did was assign my linear profile. No other differences whatever.  As you rightly point out, assigning a profile is not the same as a gamma adjustment. 

Andrew in his post with his example gave a short explanation of what's happening.