Luminous Landscape Forum
Raw & Post Processing, Printing => Colour Management => Topic started by: BrianToth on April 04, 2018, 12:53:45 pm
-
Hi, I hoping someone here can offer some suggestions on a problem I’m struggling with.
I trying to help archive hundreds of old family photos and reproduce some of them to give out to family members. They range from the 1940s to 90s. I'm not trying to restore or improve upon any of the photos, I just want them to match as close as realistically possible to the originals given the tools at my disposal. I've been having some major problems with color casts when profiling my scanners, especially when using the IT8 targets, and I've been researching and testing for three months without making any headway.
My problem is the scanning. I've tried so many combinations of things now that I've lost track.
I have an i1Studio which came with a ColorChecker 24 Mini. I first tried using this to create a profile for my Epson v800 with the i1Studio software. The profile made a noticeable improvement over the generic out-of-the-box settings, but had the effect of slightly warming up the photos a bit too much. Enough that I called X-Rite to discuss the issue. They suggested I probably needed a target with more patches like an IT8 target.
The IT8 target resulted in a very green-blue cast and muddy looking… just awful.
Long story short X-Rite was of no help, Epson support was a joke, Wolf Faust tried to help, but maybe I did’t understand enough at the time, but now I have three IT8 targets and they all result in the same color cast. Recently X-rite has unhelpfully suggested that if I’m not happy with the profiles their software makes, I don’t have to use them. Argh. (The also made me rule out scanner problems before continuing to help, so now I also have a new v600 and an old v300… they all exhibit the same results.)
As an example, here’s what I’ve been doing to test lately:
1. Scan an IT8 target, ColorChecker 24, and a family photo at the same time to guarantee the same settings. (Usually this is with VueScan RAW output, but I’ve also tried Epson Scan with no color management, and VueScan with Color Balance set to None and Output Color set to Device RGB – just to see if a gamma corrected image would profile better.)
2. Make profiles from both targets.
3. Open the original scan that includes both targets and a photo in PhotoShop and assign a profile. RGB values stay the same, Lab values change, as expected.
What I see when I sample the Lab values in Photoshop seem to corroborate what I see with my eyes, even though I know this isn’t very scientific. The Delta-E values of the profile seem to suggest they should be very good. There’s an interaction between the scans of the two targets that also seem to demonstrate the problem:
If I apply the X-rite profile from the CC24, the Lab values of the patches on the CC24 scan are pretty close to the reference values. The photo looks pretty good, though maybe a bit more yellow and pink than the original – not bad on screen, but noticeable when sitting side-by-side with the original. The grey scale in the scan of the IT8 shows a red/yellow shift rather than being closer to neutral as described in the IT8 reference file.
If I apply the X-rite profile from the IT8, the whole image gets too bright and everything shifts to the green/blue. The CC24 scan demonstrates this well: what are supposed to be neutral grey have a large color shift.
Working under the assumption that I was dealing with some sort of whitepoint/color balance issue, I researched everything I could find about ICC profiles and related issues. X-Rite’s software is a bit of a black box… I don’t know what it’s doing. So I learned how to use Argyll and wrote a script to output every combination of profile type with and without the -u and -ua flags to attempt to force an absolute rendering intent. Argyll gave some good results, but still not “correct”, and I feel like I’m just flailing around without much direction at this point. I also sidetracked into DNG profiles to try to treat the scanner as a camera… interesting experiment, but not what I need right now. I’ve also noticed that with some of the Argyll profiles that if I later convert to a standard color space using Absolute Colormetric, that the brightness shifts back to a more realistic amount. X-rite told me I should’t need to do any converting after assigning the profile, so I just don’t know…
I’m also starting to wonder if just upgrading to a target like the Digital ColorChecker SG would help me out. I feel like the CC24 does a good job, but just doesn’t have quite enough patches and the software isn’t quite extrapolating things as well as it could if it had more samples to work with. I just hesitate to throw more money at X-rite after they blew me off, but if it solves my issue I’m all for it.
Any suggestions would be appreciated. Happy to go into more details if requested, but this post was already way too long even after I skipped over a lot of things I’ve tried and researched.
I’ve uploaded my most recent test files here for anyone that want to check them out: http://server.briantoth.com/luminous/ (http://server.briantoth.com/luminous/)
Thanks so much!
-
Since i did not see it mentioned - Is your monitor profiled ?
Does a print match what you see or are you going that far yet ?
-
...
I trying to help archive hundreds of old family photos and reproduce some of them to give out to family members. They range from the 1940s to 90s. I'm not trying to restore or improve upon any of the photos, I just want them to match as close as realistically possible to the originals given the tools at my disposal. I've been having some major problems with color casts when profiling my scanners, especially when using the IT8 targets, and I've been researching and testing for three months without making any headway.
..
Brian, with respect, I really think that you are making the whole process more complex than it needs to be and would suggest that you may want to reconsider your approach.
While you may not think it you are actually trying to restore old images and in that restoration, if you actually improve then surely this must be a good thing? Looking at a B&W image from the 40's is likely to not actually appear as it did back then - it may have faded discolored etc. Similarly, with colour you will have the same issues, in fact, the issues with correction likely to be greater.
When you are dealing with old images it is my experience that you are aiming for maximum image information in the scan over the accuracy of colour as it relates to the original in its current state. Why would you want to accurately record how an image looks today after 70 years has taken its toll and changed the appearance from the way it looked originally? I believe it is better to get the best scan you can and make corrections as required to obtain the best/most pleasing image.
This may go against the grain for you and maybe others but I would suggest you ditch the colour checker, the IT8 targets and profiles they will really not help you in your quest to get best images from old B&W and Colour images. They would be of use if you actually had to colour match with best quality originals, but in this case just get in the way.
I would suggest scanning your images (crop to the image area ignoring the borders) in the required quality and in the scan software adjusting the histogram to get close to final result without clipping then in your editing application apply whatever you need to produce a pleasing image. I only looked at one image you linked to and had the briefest of plays to get what I think is a reasonable photo. Adjust the density with curves and channel mixer to remove colour, oh and a touch of dodge to the right figure to lighten a tad. Hope you are ok with me posting this? Your original scan on top my minor changes underneath
-
The Epson V800 scanner I believe also came with an entry level version of SilverFast. If that version gives you access to the SilverFast generic Reflective profile for the V800 (would be in Settings - Color Management Profile - Input), try it and see what happens. If you can't load that profile with the bundled version, download a trial copy of the upgrade to SilverFast 8 Ai which definitely allows you to colour manage the scans quite well.
-
Interestingly, I had the exact same problem of greenish color bias when calibrated with it8/2 target from Faust as discussed in this thread:
http://forum.luminous-landscape.com/index.php?topic=117912.msg976704#msg976704
Never did figure out why. Finally I gave up using calibration and eye-ball corrected the final scans.
:Niranjan.
-
The Epson V800 scanner I believe also came with an entry level version of SilverFast. …
Yes, the v800 did come bundled with SilverFast 8. (I currently have 8.8 installed.) To be honest, I haven't spent much time with it because I had purchased the i1Studio before I made a decision about which scanner to purchase. And when I ran into my initial color issues with the IT8 targets, X-Rite and B&H figured there must have been something wrong with the v800 because it was "supposed to make excellent scans right out of the box". The bundled SilverFast wouldn't support my v300 for comparison so I resorted to just using Epson Scan and VueScan for all future tests. I also later purchased a v600, which wouldn't work with the bundled SilverFast either of course. Also, my copy of SilverFast won't scan to 48bit mode unless I put it in 48bit HDR mode.
That said, I did just do a test. (I hope I did it correctly.) In SilverFast I went to Preferences > CMS and chose 'SFprofR (Perfection v800)' as the Input profile. I did a scan with every other option set to the default (e.g. '0'), in 48-24bit mode, and I got a TIFF with an embedded 'SFprofR (Perfection v800)' profile.
(I just noticed though that if I go back to Preferences > CMS I see that it didn't retain the chosen Input profile, so it probably wants me to choose an Internal profile as well. I was hoping to be able to compare the pure scanner profiles without doing any sort of transform, but even with the Internal profile set the Lab values between to two are pretty much the same… hmm.)
Anyway, the scan still exhibits a green/blue shift in the photo and the ColorChecker, but not as much as the X-Rite or Argyll IT8-based profiles. I can't do my own SilverFast profile without upgrading.
I uploaded those two scans to here (http://server.briantoth.com/luminous/SilverFast%20Scan%20+%20SilverFast%20Profile/).
Something I haven't been quite clear on: is it normal to have to adjust white balance on a scanned image after assigning a profile? I would think 'no', but maybe I'm just missing something simple?
-
Of the two scans, the one that is not sRGB looks more natural (less saturation) and more neutral. There seems to me to exist a slight magenta cast in the frame of the Faust target, but the ColorChecker and the B&W photo look not too bad. For a comparison, I just did for you a scan of my ColorChecker using the canned SilverFast reflective profile for the V800 (same as for the V850 I'm using) and other settings in the CMS as attached. If you are on Windows, "ColorSync should say ICM. The result is also attached. It should look pretty neutral. A TIFF I made of it could be more neutral and more consistently neutral in the grayscale patches, but this is an acceptable starting point amenable to marginal improvement with a bit of tweaking in a post scan editing application or SilverFast itself.
-
Yes, the v800 did come bundled with SilverFast 8. (I currently have 8.8 installed.) To be honest, I haven't spent much time with it because I had purchased the i1Studio before I made a decision about which scanner to purchase. And when I ran into my initial color issues with the IT8 targets, X-Rite and B&H figured there must have been something wrong with the v800 because it was "supposed to make excellent scans right out of the box". The bundled SilverFast wouldn't support my v300 for comparison so I resorted to just using Epson Scan and VueScan for all future tests. I also later purchased a v600, which wouldn't work with the bundled SilverFast either of course. Also, my copy of SilverFast won't scan to 48bit mode unless I put it in 48bit HDR mode.
That said, I did just do a test. (I hope I did it correctly.) In SilverFast I went to Preferences > CMS and chose 'SFprofR (Perfection v800)' as the Input profile. I did a scan with every other option set to the default (e.g. '0'), in 48-24bit mode, and I got a TIFF with an embedded 'SFprofR (Perfection v800)' profile.
(I just noticed though that if I go back to Preferences > CMS I see that it didn't retain the chosen Input profile, so it probably wants me to choose an Internal profile as well. I was hoping to be able to compare the pure scanner profiles without doing any sort of transform, but even with the Internal profile set the Lab values between to two are pretty much the same… hmm.)
Anyway, the scan still exhibits a green/blue shift in the photo and the ColorChecker, but not as much as the X-Rite or Argyll IT8-based profiles. I can't do my own SilverFast profile without upgrading.
I uploaded those two scans to here (http://server.briantoth.com/luminous/SilverFast%20Scan%20+%20SilverFast%20Profile/).
Something I haven't been quite clear on: is it normal to have to adjust white balance on a scanned image after assigning a profile? I would think 'no', but maybe I'm just missing something simple?
The second scan you made doesn't look too bad at all. The color checker looks pretty good. I think you are just seeing the color of the old photographic paper from the print, and perhaps some UV effect from optical brighteners in the printed photograph.
I think you're good to go!
-
Of the two scans, the one that is not sRGB looks more natural (less saturation) and more neutral.
Thanks for the sample. Interesting to see the advanced profile options in the higher up version of SilverFast.
I have a question though… my understanding is that the greys of the ColorChecker should be close to neutral, with perfect neutral meaning that R=G=B or in Lab that a*=b*=0. For example, the X-Rite reference file for the ColorChecker shows patch D4 to supposedly have Lab values around 50.76, -0.13, 0.14. Patch D4 in the scan shows an RGB of 83, 94, 96 and Lab of 38.53, -6.335, -3.343.
While the photo does look acceptable with any of the profiles on its own, with the IT8-based profiles (or the built-in profiles) assigned it's very green/blue compared to the original. That's hard to judge without having the original side-by-side, so I was trying to go by the target patches as a less subjective measurement.
-
Thanks for the sample. Interesting to see the advanced profile options in the higher up version of SilverFast.
I have a question though… my understanding is that the greys of the ColorChecker should be close to neutral, with perfect neutral meaning that R=G=B or in Lab that a*=b*=0. For example, the X-Rite reference file for the ColorChecker shows patch D4 to supposedly have Lab values around 50.76, -0.13, 0.14. Patch D4 in the scan shows an RGB of 83, 94, 96 and Lab of 38.53, -6.335, -3.343.
While the photo does look acceptable with any of the profiles on its own, with the IT8-based profiles (or the built-in profiles) assigned it's very green/blue compared to the original. That's hard to judge without having the original side-by-side, so I was trying to go by the target patches as a less subjective measurement.
Hi Brian,
In principle you are correct that the gray patches should measure close to their file reference values. In the sample scan I made there are deviations, which is why I said "A TIFF I made of it could be more neutral and more consistently neutral in the grayscale patches, but this is an acceptable starting point....." above. There are four reasons for it: (1) My CC card is about 15 years old; while kept in dark storage some of its colours could have shifted a bit over the years. X-Rite recommends replacing it every few years. (2) I used a canned profile rather than a custom profile; it's possible my V850 doesn't behave exactly like the one LaserSoft Imaging used for creating their packaged profile. (3) The quality of the target influences the quality of the profile, so it could be that superior quality targets would produce custom profiles that show better neutrality characteristics; (4) Producing exact colour reproduction from a scan is generally hard to achieve; for example, I didn't let the lamp warm up (not supposed to be necessary with this model, but who knows), lamp temperature can shift overtime, scanner hardware less than perfect; etc. Part of this issue is to align ones expectations with technical reality. Notwithstanding all I mentioned here, the very "quick and dirty" result I produced with a canned profile, as I said, is a good starting point for refinement in post-scan workflow.
In what I quoted from you above, I simply don't understand the sentence I put in bold. Please explain.
-
I have 3 full size CCs. An old GMB one from around 2004, an X-Rite one from 2009, and an X-Rite "Classic" from 2017.
All the neutral patches are within 1 dE of each other and the a*, b* are within < 1 from neutral except the white patch which is slightly warm. b* between 2 and 3.
The sort of shift seen here is not from CC changes. The neutrals are the most stable patches over time. It's been a while since I profiled a scanner but I never saw anything like these variations. Something is wrong in the process. These differences are way too large.
-
Referring to Doug's reply on the scan values:
The reasons I listed above are just generic possibilities. The least likely culprit is the CC card, as you say. I just checked mine with my i1Pro2 and it's good - if it wandered at all, it's really minor. i.e. the L* values are within no more about 2 dE from where they are supposed to be, and the a* b* values are on the whole less than 1 dE from 0 save for two readings - the worst 2.76.
Forget about the example I put up yesterday - it wasn't a designed accuracy test. I just wanted to get a quick impression of whether it would produce visibly noxious results scanning the way one would normally scan every day photos, so I adjusted the tails of the histogram, scanned it in ARGB(98) and then converted it from TIFF to JPEG. So it's been mangled from a perspective aimed at proofing the accuracy of the scan workflow. It needed a different kind of scan set-up to assess accuracy of the profile and the scanner. But that's not all that's "wrong", because having produced a result with a group of settings that produce results as faithful to the values on the CC card as I could manage it with the V850, there are still issues with the deeper grays, but much less serious than yesterday's. In particular, apart from the changed CMS settings, no histogram adjustments in the "Unadjusted" exercise.
The attachments pretty much tell the story, so I won't go into muchdescription here. Anyone who needs clarification or further explanation, please ask and I'll do so. Re the panels of numbers, perhaps I should just explain up-front: the "ADJUSTED" set are yesterday's settings provided yesterday. The "UNADJUSTED" set underneath are today's settings, selected to produce the closest scan to the original card from scanner through SilverFast through opening in Photoshop, all attached to this post.
What I find most troublesome are the "differences of the differences" as between cols, HIJ involving eyedropper readings from the scan opened in Photoshop versus columns NOP involving readings with SilverFast's densitometer of the scan open in SilverFast, using nominally the same internal working space profile and Absolute Rendering Intent in both applications.
Sorry forgot to mention - this is still the canned profile. I should repeat the scan with the same settings and my custom profile to see what improves.
-
I have 3 full size CCs. An old GMB one from around 2004, an X-Rite one from 2009, and an X-Rite "Classic" from 2017.
All the neutral patches are within 1 dE of each other and the a*, b* are within < 1 from neutral except the white patch which is slightly warm. b* between 2 and 3.
The sort of shift seen here is not from CC changes. The neutrals are the most stable patches over time. It's been a while since I profiled a scanner but I never saw anything like these variations. Something is wrong in the process. These differences are way too large.
Hi Doug,
Thanks for this info, that's very good to know. Do you have any experience with the Digital ColorChecker SG?
When I make profiles using the ColorChecker 24, the patches of the scanned target do match the reference values, which is what I would expect. But when applied to the photos they end up a bit too warm. It was suggested to me by X-Rite, and I've read from others, that the 24 patches are probably not really enough for a great scanner profile. Wondering of the Digital ColorChecker SG would be a better solution given it's made with a similar manufacturing process to the CC24, but with whiter whites, blacker blacks, and many more patches.
The drastic color shift only seems to be a product of the IT8 profiling or when using the manufacturer scanner profiles. Obviously I'm curious why, but if the Digital ColorChecker SG bypasses the problem, that'd be easier. :)
-
Hi Doug,
Thanks for this info, that's very good to know. Do you have any experience with the Digital ColorChecker SG?
When I make profiles using the ColorChecker 24, the patches of the scanned target do match the reference values, which is what I would expect. But when applied to the photos they end up a bit too warm. It was suggested to me by X-Rite, and I've read from others, that the 24 patches are probably not really enough for a great scanner profile. Wondering of the Digital ColorChecker SG would be a better solution given it's made with a similar manufacturing process than the CC24, but with whiter whites, blacker blacks, and many more patches.
The drastic color shift only seems to be a product of the IT8 profiling or when using the manufacturer scanner profiles. Obviously I'm curious why, but if the Digital ColorChecker SG bypasses the problem, that'd be easier. :)
The CC SG is good for making profiles as it has a surprising large number of patches that are rather saturated. They also are mostly the darker ones that you might not guess are that saturated. Many way outside of sRGB unlike the standard CC where only the cyan patch exceeds sRGB.
That said, using the CC SG requires much more care. Because it has a semigloss surface (hence offers more saturated colors), it is very susceptible to specular reflections from the environment. My use of it in the past was to profile cameras for reproduction purposes (scene referred). To deal with the specular reflections I would use a large, black umbrella with a hole cut in it for the camera lens. This in addition to illuminating it evenly at 45 degrees. That setup most closely replicates the geometry that the patches are measured with a spectrophotometer.
I don't know whether the SG surface is an issue with scanners. However, it does have a larger gamut along with more colors. I haven't done any critical work in a long time with scanners.
I do find your problems with the IT8 odd. OTOH, my workflow with both camera and back in the day, scanner profiles, is somewhat different. I assign, then convert to Adobe RGB, the generated profiles in Photoshop to images processed the same way as the targets used to create the profiles. While somewhat unorthodox, I understand how and why that works. I have never tried the IT8 approach Vuescan incorporates though I have used Vuescan. For one thing I find their description opaque. I want to understand the math of what I do.
-
So, While the photo does look acceptable with any of the profiles on its own, with the IT8-based profiles (or the built-in profiles) assigned it's very green/blue compared to the original.
In what I quoted from you above, I simply don't understand the sentence I put in bold. Please explain.
Hi Mark,
Sure, I worded that horribly. :D
If I assign my IT8-generated profiles to the scans… they do look better than the "RAW" image that came out of the scanner. Same for scans made with either the Epson Scan-supplied profile, or the SilverFast Epson profile. So without anything to compare them too, I'd say they're good and look fine. But when compared to the original (especially when printed out and compared under the same lighting) they're noticeably color-shifted. This is hard to demonstrate though because obviously I don't have a way to accurately share what the original looks like, and I'm not very good at explaining colors anyway. So that's why I was trying to go by the target patches as a less subjective measurement. Numbers are easier for me to explain and understand. But I've also read that the numbers might not always tell the whole story with this stuff. :)
The exception to this is my profile made with the CC24. If I assign that profile, then the CC24 patches match the reference file almost spot-on, but other scanned items end up a bit too warm. This may simply be a limitation of only having 24 patches to work with though and perhaps a different target would solve my issue in that regard.
Hope that makes more sense.
-
Forget about the example I put up yesterday - it wasn't a designed accuracy test. …
No, thanks for taking the time to provide the examples! Both of your posts above have actually been very informative as I don't have access to v850 to test. And it looks like your results are actually very similar to what I'm seeing with my equipment and that is actually good news to me. I might have some more thoughts/questions about your results after I process all the info you shared.
Sorry forgot to mention - this is still the canned profile. I should repeat the scan with the same settings and my custom profile to see what improves.
If you do happen to attempt it with your own custom profile, I'd be very curious to see the results. I'm assuming based on other posts I've come across of yours that you used an IT8 with the built-in SilverFast AI profiling? When I first started running into this problem I wondered if I would've had better results with the v850 kit vs my own profiling.
-
Brian,
V800 and V850 - same principles of colour management apply, so if you are using a V800, everything I say about colour managing a V850 is applicable.
Let's focus mainly on three things:
(1) Making profiles
(2) Selection of where to profile the scanner
(3) Selection of colour working space ("internal" profile)
(1) Making profiles: The SilverFast Auto IT8 process is the easiest, most practical approach I've ever worked with, and the results are generally very usable. I've tested the X-Rite approach (see my article on the Epson V850 scanner, found it convoluted and cumbersome and no more accurate than SilverFast Auto IT8. To use it, however, you need the LaserSoft Imaging IT8 targets, because it involves a bar code on the target. There are not enough patches in a CC24 to make adequate scanner profiles. The CCSG was not designed for scanner profiling. It is best to use bespoke scanner profiling targets.
(2) Selection of where to profile the scanner: you have essentially two choices: (i) in the scanning process, or (ii) in Photoshop after the scan is made. For option (i) you select ColorSync/ICM in the upper section Color Management, Input to working space. Then in the middle section - Profiles - for input you select your scanner profile, whether a custom one or the supplied canned one. For option (ii), instead of ColorSync/ICM in the Color Management Section Input to Internal you select <none> because you will not be colour managing in SilverFast. If you select <none> here, then there will be no choice of a scanner profile for Input in the Profiles section; rather you will assign the scanner profile to the photo when you open it in Photoshop and then convert the image to the preferred colour working space (ARGB(98), ProPhoto etc.).
(3) Selection of Color Working Space (the "internal" profile): We generally recommend using a wide space because scanner colour spaces tend to be wide and if the scan will be sent to a modern inkjet printer, and especially using luster/gloss papers for printing, the gamut can be wider than aRGB(98). So ProPhoto is a good choice provided you make the scans in 16-bit to avoid banding that can happen if there are too few levels for the size of the colour space.
Not mentioned above, select the Rendering Intent that makes the image look best to you. This manages how out-of-gamut colours are handled.
For the monitor, you want it to be properly calibrated and profiled and ColorSync to be active in SilverFast CMS Color Management section, "Working Space to Monitor". SilverFast will then use your active monitor profile.
-
I have 3 full size CCs. An old GMB one from around 2004, an X-Rite one from 2009, and an X-Rite "Classic" from 2017.
Pffft, real men use color checkers from the '70s.
-
Hi, Brian:
It is a little too much for me to keep abreast of this thread so I may be wrong to surmise this generally: when you create a profile using Faust's it8 target, you get the luminosity values in the ball park but the a, b values show green/blue shift. If this is true, and since that matches my experience exactly on a different scanner (Epson 3200,) perhaps time to question the quality of Faust's targets. Given that they are so cheap, may we get what we pay for.
:Niranjan.
-
Hi, Brian:
It is a little too much for me to keep abreast of this thread so I may be wrong to surmise this generally: when you create a profile using Faust's it8 target, you get the luminosity values in the ball park but the a, b values show green/blue shift. If this is generally true, and since that matches my experience exactly on a different scanner (Epson 3200,) perhaps time to question the quality of Faust's targets. Given that they are so cheap, may we get what we pay for.
:Niranjan.
I wouldn't jump to any conclusions on the quality of Faust targets. There are a number of variables at play here and unless you can isolate the particular influence of each you can't reliably pinpoint the culprit or culprits.
-
I wouldn't jump to any conclusions on the quality of Faust targets. There are a number of variables at play here and unless you can isolate the particular influence of each you can't reliably pinpoint the culprit or culprits.
I understand. If that was not true, it is an interesting coincidence to see the same outcome in two different independent cases. We both could have been doing same thing (or different) erroneous in the process, I guess. May be something amiss in the way VueScan does the number-crunching. We can learn where the problem lies, the process or the target by having one of us send someone else the target and have them do calibration with the a known good process and see what results.
:Niranjan.
-
Sure, but as you most likely know, coincidences are not necessarily determinative. Nor is it necessarily the case that both of you were doing the same thing in the same wrong way - your procedures may have been satisfactory but the problem is elsewhere. Based on what we have seen so far, it's largely speculative. To add to the speculation, I wouldn't rule out the possibility of technical limitations to the accuracy of reflective scanning with these scanner models. If you look at the numbers I presented yesterday - here with an Epson V850, using SilverFast Ai 8 Studio and the SilverFast target, check what happened in the panel bounded by cols H-I-J rows 10 to 17. The performance of the L* and b* channels is not too bad, but that of the a* channel deteriorates as the grayscale patches get darker. One wonders whether this could be evidence of scanner noise having increasing impact on accuracy as the signal:noise ratio decreases with lower luminance. Nothing conclusive, but again, another possibility to consider. And I remain concerned about some of the measurements themselves - still bothered by the inconsistency of readings between SilverFast and Photoshop, in particular comparing the a* channel col L (SilverFast) versus col C (Photoshop), rows 10 to 17.
-
I wouldn't jump to any conclusions on the quality of Faust targets. There are a number of variables at play here and unless you can isolate the particular influence of each you can't reliably pinpoint the culprit or culprits.
I just happened to purchase and receive my Faust target a couple days ago and used it to profile my Nikon Coolscan V ED under Vuescan.
The process was very simple and it appears dead-nuts accurate from the results I've been getting.
Allan
-
I understand. If that was not true, it is an interesting coincidence to see the same outcome in two different independent cases. We both could have been doing same thing (or different) erroneous in the process, I guess. May be something amiss in the way VueScan does the number-crunching. We can learn where the problem lies, the process or the target by having one of us send someone else the target and have them do calibration with the a known good process and see what results.
:Niranjan.
Interesting that you had some of your own issues. I probably came across your discussion during all my research.
For what it's worth, the reason I ended up with three IT8 targets is that I initially suspected a problem with the first one. That first one was $60 from LaserSoft (i.e. SilverFast). I then bought two more from Wolf Faust ($10 ea). One on Kodak, the other on Fuji. All three exhibit the same problem. For all I know, Wolf Faust manufactures the LaserSoft ones. :) Either way, they all seem to perform the same, and according to my profiling results, the Delta-E values should be well within acceptable margins of error… which I would take to imply that there's no noticeable discrepancies between the target itself and its reference file.
I've also tried the following scan methods (for both generating the profile and future scans that are to use the profile):
- Epson Scan with "no color management" (turns off all settings). Output is still gamma(?) corrected or internally assigned some standard profile, but no profile is saved with the image.
- VueScan "RAW". Gamma 1.0, no corrections. (X-Rite doesn't like to make profiles from these, so mostly tried them with Argyll CMS.)
- VueScan with "Color: None", default values, output to Device RGB.
- SilverFast "RAW"
All methods produced pretty much the same results once the profiles were assigned to the image. Some had slightly better shadow details, etc. but without getting too fussy, they all had the same color shifts.
Also tried with a v800, v600, v300. The v300 has very noisy blacks, but otherwise they all look very similar.
Using Argyll CMS from the command line gave me a ton more control but the cLut profiles give the same color shifts. Some like the Shaper Curve profiles look pretty good on the screen, but that's a whole other can of worms… too many options and I don't know which I'm supposed to be focusing on. I also don't know what type X-Rite produces to compare with. Maybe there's a way to figure it out using the ColorSync utility?
I attached a PDF of a small sampling of the Lab values comparing the four profile results.
-
I just happened to purchase and receive my Faust target a couple days ago and used it to profile my Nikon Coolscan V ED under Vuescan.
The process was very simple and it appears dead-nuts accurate from the results I've been getting.
Allan
I've tried using the VueScan profiling. For reflective scanning at least, I didn't find it any more useful than using the built-in device profile. It doesn't help at all with gamma correction, white balance, etc. which still leaves me a lot of guesswork to get acceptable results. Since you're using a Nikon Coolscan V ED, I'm assuming you had to use a transparent IT8 target to do your profiling? Is there any way that you could compare a scan of your IT8 with the profile applied to the IT8 reference Lab values? I'm not doing any slide scanning yet, but I do have some in my collection that I might need to tackle some day.
-
Either way, they all seem to perform the same, and according to my profiling results, the Delta-E values should be well within acceptable margins of error… which I would take to imply that there's no noticeable discrepancies between the target itself and its reference file.
If you are getting the same results with a range of different charts and profiling packages, then that points to either a workflow issue outside the profiling process, or a spectral mismatch between the media you are scanning and the charts. Lacking a spectral characterization of your scanner, the only way of guaranteeing optimal results using test charts and tri-stimulus profiling is to have test charts based on identical paper and colorants to the media you wish to scan.
-
If you are getting the same results with a range of different charts and profiling packages, then that points to either a workflow issue outside the profiling process, or a spectral mismatch between the media you are scanning and the charts. Lacking a spectral characterization of your scanner, the only way of guaranteeing optimal results using test charts and tri-stimulus profiling is to have test charts based on identical paper and colorants to the media you wish to scan.
Exactly so. A colorimetric (accurate) scan isn't possible unless the scanner's illuminant is something close to D50. However, without a D50 illuminant and color filters that mean L/I, one can accurately scan an image printed on the same printer that printed the scanner target. It just likely will be off scanning other media with different spectral characteristics. The same applies using a scanning target printed with differing spectral responses.
Exactly the same problem occurs when profiling cameras for scene referred repro work. You just hope to get close enough.
-
If you are getting the same results with a range of different charts and profiling packages, then that points to either a workflow issue outside the profiling process, or a spectral mismatch between the media you are scanning and the charts. Lacking a spectral characterization of your scanner, the only way of guaranteeing optimal results using test charts and tri-stimulus profiling is to have test charts based on identical paper and colorants to the media you wish to scan.
Taking things one step at a time, at the very least round-tripping the scanner target should produce a scan whose read values are close to the reference values, unless there are issues. Starting with that, one can determine whether there are issues (hardware, software, settings) before taking on media that differs.
-
Taking things one step at a time, at the very least round-tripping the scanner target should produce a scan whose read values are close to the reference values, unless there are issues. Starting with that, one can determine whether there are issues (hardware, software, settings) before taking on media that differs.
Right. Highly recommended.
-
If you are getting the same results with a range of different charts and profiling packages, then that points to either a workflow issue outside the profiling process, or a spectral mismatch between the media you are scanning and the charts. Lacking a spectral characterization of your scanner, the only way of guaranteeing optimal results using test charts and tri-stimulus profiling is to have test charts based on identical paper and colorants to the media you wish to scan.
Exactly so. A colorimetric (accurate) scan isn't possible unless the scanner's illuminant is something close to D50. However, without a D50 illuminant and color filters that mean L/I, one can accurately scan an image printed on the same printer that printed the scanner target. It just likely will be off scanning other media with different spectral characteristics. The same applies using a scanning target printed with differing spectral responses.
Exactly the same problem occurs when profiling cameras for scene referred repro work. You just hope to get close enough.
Thanks for the feedback. I had a hunch that might be playing a role to some extent. I've read about the importance of having the target match the materials I'm scanning, though I had also read it was more critical for some media than others.
“Close enough” would be good. The color shift happens across all kinds of photos that I'm scanning, from very glossy to matte, color or black & white.
It looks like most of the photos I'm scanning have much whiter paper than the IT8 target white patches, which is maybe why the ColorChecker 24 handles the white point better? Yet the black patch on the ColorChecker does't get anywhere near as black as the black and white photos I have.
Do you think my next best option would be to use a target that has a wider gamut that might encompass the media I am scanning better?
When I veered off into making DNG camera profiles as an experiment, the Adobe DNG Profile Editor suggested that the color temperature of the scanned ColorChecker was 6800k. I'm guessing it just estimated that based on the white balance correction that it applied to the profile, but that was something I tried to learn about… how that might affect the whitepoint of the ICC profiles.
I can try to do some test readings using Argyll and my i1Studio spectrophotometer if that would help with anything.
-
Taking things one step at a time, at the very least round-tripping the scanner target should produce a scan whose read values are close to the reference values, unless there are issues. Starting with that, one can determine whether there are issues (hardware, software, settings) before taking on media that differs.
I think the targets do an ok job of profiling themselves. Especially the ColorChecker. The IT8 profiles raise the luminance value too much though, unless I use the -u or -ua flags in Argyll to make the profile. Alternatively assigning the IT8 profile and then converting to the working space using absolute colorimetric brings the values closer to the reference. Seems like a roundabout way to do it, but I’m sure I’m just not understanding the internal workings of the profiles very well, at least when it comes to input profiles. Profiles made from the ColorChecker seem to do better in this regard.
Is there a recommended tool I could use to generate automated comparisons between the reference data and the profiles in a meaningful way? Right now I have to manually sample each patch in Photoshop, which is way too time consuming. I have the target reference files, scans of the target, and if I use Argyll I have .ti3 files from scanin. I attached the results of Argyll’s profcheck against some of my profiles, as well as a sample of manually read Lab values from Photoshop, but it’d be great if I could script some way to mass-compare profiles somehow. Going just off Delta-E seems to not mean much, sometimes the smaller error values actually yield greater color shifts than the simpler profiles.
-
Is there a recommended tool I could use to generate automated comparisons between the reference data and the profiles in a meaningful way?
Chromix ColorThink Pro.
-
I did get a Digital ColorChecker SG. At the moment, I haven't seen any advantage to it. According to ColorThink (which is super buggy), the gamut volume of the SG profiles are less than that of the IT8, which isn't what I was expecting… considering the SG colors are more saturated. I did attach some preliminary ColorThink info. (I don't own a licensed copy and considering how much it's crashed, I think I'll just do things the old fashioned way.)
I found this interesting paragraph on the ICC's website (http://www.color.org/faqs.xalter#pa1):
Q. How do I assess the accuracy of an input profile?
A. Make sure you have a CIELAB colour space profile on your system, and select an input test chart such as the one you made the profile with (and which you have measurement data for. Then use Photoshop 'convert to profile' command to convert from your input profile to the CIELAB colour space, using the absolute colorimetric intent and no black point compensation. The CIELAB values (which you can display as a CIELAB image in Photoshop) should match the measurement data for the chart. Note that you are only evaluating the colorimetric rendering intent by this method, and not the perceptual rendering intent.
It hadn't occurred to me to do this. I was comparing the LAB values after assigning my profile to the scanned image. In hindsight it makes sense as most of the profiles use the target white patch as the whitepoint, which doesn't make sense when trying to compare to the target reference file. Also, as I had already observed, when I would convert to my working space using different intents, it would sometimes bring the luminosity values back in line with the target reference file.
None of this hasn't magically solved any of my problems. But I wonder what other detail I might be missing that's leading to my poor scan results. I suspect I've had a misunderstanding of how and when to manage the white point throughout the scanning > display > printing process. I may have been trying to force an absolute intent to get the values to match, but what I really might want is to keep the white point relative to the media I'm viewing the image on. I don't know.
-
ColorThink Pro is not buggy at all.
A ColorChecker SG is not meant for profiling scanners. Don't use it. You need proper scanner profiling targets. ColorThink Pro Color Worksheets will give you a reliable reading of the performance of properly made and correctly scanned targets.
-
A ColorChecker SG is not meant for profiling scanners.
Sure you can use a ColorChecker SG for profiling scanners, if it better (spectrally) represents the type of media you want to scan. For artwork people have gone to the trouble of making custom test charts using the exact canvases and paints of the type of work they want to scan, and if that is the type of thing you are using your scanner for, then a ColorChecker is going to be closer to the mark than a photographic IT8 print.
-
Sure you can use a ColorChecker SG for profiling scanners, if it better (spectrally) represents the type of media you want to scan. For artwork people have gone to the trouble of making custom test charts using the exact canvases and paints of the type of work they want to scan, and if that is the type of thing you are using your scanner for, then a ColorChecker is going to be closer to the mark than a photographic IT8 print.
As I see it the main issue with both ColorChecker and Colorchecker CG is whether the scanner is impacted by the air gap between the color patches and glass since they are slightly recessed. My sense is that this depends on the scanner.
I got better scene referred profiles from printed colorcheckers than an actual colorchecker. I believe this is because the printed colorchecker spectral content is spectrally less diverse than an actual colorchecker. Thus a matrix profile fits quite well.
So I have used profiles from a printed colorchecker for repro work of printed material that is CYMK and LUT profiles from a regular ColorChecker and/or SG for items that are more diverse spectrally.
But I don't have any hard evidence one way or the other. Just very limited data sets and the scanner I used was retired some years ago.
-
Sure you can use a ColorChecker SG for profiling scanners, if it better (spectrally) represents the type of media you want to scan. For artwork people have gone to the trouble of making custom test charts using the exact canvases and paints of the type of work they want to scan, and if that is the type of thing you are using your scanner for, then a ColorChecker is going to be closer to the mark than a photographic IT8 print.
If you have evidence of a scanner profile made from a CCSG being superior (for the vast majority of photographs people scan) to say the Hutchcolor scanner profiling targets, or the Faust targets, or the LSI targets, that would be interesting to see. As for targets bespoke to certain kinds of artwork, judging from all I've been exposed to, to get results those folks consider satisfactory the selection of the palettes for such profiles seem a lot more painstaking and customized than just using a CC.
-
I maybe be naïve, but how can scanners be expected to be profiled into yielding accurate color? For example, here is the light in my Epson V600 scanner:
(http://www.fishcreekstudio.com/timages/01/spectrum-epson-v600.png)
If this can be profiled into giving accurate color, then why are we wasting time searching for lights that provide accurate color rendering? Why don't we just grab whatever LED bulbs are in our local hardware store and profile our cameras just like scanners are profiled?
Sorry if this is an elementary question but I've had cognitive dissonance ever since I measured my V600's LED.
-
I maybe be naïve, but how can scanners be expected to be profiled into yielding accurate color? For example, here is the light in my Epson V600 scanner:
(http://www.fishcreekstudio.com/timages/01/spectrum-epson-v600.png)
If this can be profiled into giving accurate color, then why are we wasting time searching for lights that provide accurate color rendering? Why don't we just grab whatever LED bulbs are in our local hardware store and profile our cameras just like scanners are profiled?
Sorry if this is an elementary question but I've had cognitive dissonance ever since I measured my V600's LED.
It's a problem, but not as big as one might think.
Take a look at the spectrum of a white patch on your monitor. Pretty peaky. But can it be profiled reasonably accurately? Yes. So it's not the issue of an illuminant alone but the interaction of the illuminant with the spectral diversity of objects being illuminated.
This is why a matching print of a colorchecker can be profiled with a lower dE than an actual colorchecker. This is also why using a scanner to scan photos can yield good results. Both photos (color dyes) and prints (from CYMK) printers have less diversity than things like colorcheckers which are made from spot colors from a variety of pigments. However, both photos and prints need separate profiles made from patch sets using the two processes for optimal scanning color accuracy.
For other objects scans will introduce more color distortion.
-
I always wondered same questions
-
The uselessness of scanner profiles
I figured it was time to circle back to this thread. This was my first posting to this forum, out of frustration and desperation after months of trying to figure out what I was doing wrong. Then I took a step back and made sure my fundamentals were correct before continuing on. For anyone that's interested, I took what I learned and all my trial and error techniques and finally put together a concise analysis of some of my scanner profiles. I'm sure to the pros out there that this is nothing new, but I found it to be an extremely clear way of showing the problem I'm facing and might help some other amateur like me not waste too much of their time.
Applying an ICC profile back to the scanned image that created the profile
Let's start with the classic IT8 target that's bundled with a lot of higher-up scanner packages. I have three of them, one from LaserSoft, two from Wolf Faust (one Fuji, one on Kodak). They all give me the same results FYI. This example uses one from Wolf:
(http://server.briantoth.com/luminous/colorthink/small/IT8%20+%20Xrite%20IT8%20ICC.png) (http://server.briantoth.com/luminous/colorthink/fullsize/IT8%20+%20Xrite%20IT8%20ICC.png)
(Click for larger image.) On the left are the RGB values of the scanned target using Epson Scan with No Color Management. (I sampled the RGB values from the scanned image as a color list.) I created an ICC profile using i1Studio and applied it to the RGB values with an absolute rendering intent, resulting in the LAB values in the next column. Moving on are the LAB values from the IT8 reference file provided with the target. The final column are the dE2000 values between the profiled image and the reference values. Green are <= 1dE, yellow up to 2.5dE, etc.
Here's the similar results from a ColorChecker 24 and a Digital ColorChecker SG. (Note that I made my own reference files because there were a few significant deviations between my targets and the bundled i1Studio reference files.)
(http://server.briantoth.com/luminous/colorthink/small/CC24%20+%20Xrite%20Custom%20CC24%20ICC.png) (http://server.briantoth.com/luminous/colorthink/fullsize/CC24%20+%20Xrite%20Custom%20CC24%20ICC.png)(http://server.briantoth.com/luminous/colorthink/small/DCCSG%20+%20Xrite%20Custom%20DCCSG%20ICC.png) (http://server.briantoth.com/luminous/colorthink/fullsize/DCCSG%20+%20Xrite%20Custom%20DCCSG%20ICC.png)
All this does is confirm that the profiling software can correct for the scanned target itself, but that's something. Presumably, that means my equipment and software are working and that my workflow isn't just wrong.
Applying a profile from one target to another
This basically fails. I'll share two that I thought were interesting because they kinda backup my original observations with the color shifts.
First, a scanned IT8 with a profile from the DCCSG. There are obvious color shifts in the greyscale patches.
(http://server.briantoth.com/luminous/colorthink/small/IT8%20+%20Xrite%20Custom%20DCCSG%20ICC%20(Grey%20Scale%20Shift).png) (http://server.briantoth.com/luminous/colorthink/fullsize/IT8%20+%20Xrite%20Custom%20DCCSG%20ICC%20(Grey%20Scale%20Shift).png)
Second, the DCCSG with the IT8 profile. It's harder to see but the greys and blacks took on a greenish cast. The LAB values would seem to confirm it. Plus the dE values are way out there anyway.
(http://server.briantoth.com/luminous/colorthink/small/DCCSG%20+%20Xrite%20IT8%20ICC.png) (http://server.briantoth.com/luminous/colorthink/fullsize/DCCSG%20+%20Xrite%20IT8%20ICC.png)
So, I'm sure in hindsight this should be obvious. All three targets I'm working with have different spectral characteristics. This leads to my prints…
Applying the profiles to a printed photo
I took an image of the ColorChecker 24 from BabelColor that was already in LAB colorspace and printed it using an absolute intent to a custom paper profile also made with the i1Studio. I measured the print with my i1Studio spectrophotometer and did the same tests. The only profile that did a good job was a custom target of 962 patches that I made with Argyll on the same paper. The other targets weren't good enough.
The ColorChecker profile couldn't even correct the printed ColorChecker:
(http://server.briantoth.com/luminous/colorthink/small/Printed%20CC24%20+%20Xrite%20Custom%20CC24%20ICC.png) (http://server.briantoth.com/luminous/colorthink/fullsize/Printed%20CC24%20+%20Xrite%20Custom%20CC24%20ICC.png)
Custom Argyll target:
(http://server.briantoth.com/luminous/colorthink/small/Printed%20CC24%20+%20Argyll%20Custom%20Target%20XYZ%20cLUT%20ICC.png) (http://server.briantoth.com/luminous/colorthink/fullsize/Printed%20CC24%20+%20Argyll%20Custom%20Target%20XYZ%20cLUT%20ICC.png)
For all I know this could be even better if I made sure to have the same print settings.
Conclusion of the testing
So this seems to prove what was said before by Doug, Graeme, and others that are much more knowledgeable about this stuff. But I'm happy that I proved it to myself. That said, now I'm kinda annoyed. Scanner profiling is marketed as a simple solution to getting your colors right and it's clearly not. I'd feel ripped off if it weren't for the fact that the i1Studio has worked for print profiling and display profiling. Glad I didn't spend big bucks on third party scanner profiling software. I did spend a ton of time with Argyll and got similar results to the i1Studio in the end. But Argyll is awesome and I couldn't have done these tests without the other tools it provides like its spotread tool – because the i1Studio doesn't offer that capability. You can see all the test results here (http://server.briantoth.com/luminous/colorthink/fullsize/).
-
I maybe be naïve, but how can scanners be expected to be profiled into yielding accurate color? For example, here is the light in my Epson V600 scanner:
(http://www.fishcreekstudio.com/timages/01/spectrum-epson-v600.png)
If this can be profiled into giving accurate color, then why are we wasting time searching for lights that provide accurate color rendering? Why don't we just grab whatever LED bulbs are in our local hardware store and profile our cameras just like scanners are profiled?
Sorry if this is an elementary question but I've had cognitive dissonance ever since I measured my V600's LED.
I forgot that you posted this, but for the last few months my assumption was that the color shifts I was seeing had to be influenced somehow by the LEDs. Doug made a post regarding the flare of his v850 affecting his profiles, and other topics I read seemed to indicate this could be an issue.
The past couple weeks I've been reading about going the copy stand route of photographing my old photos. I started looking into lighting solutions and came across discussions of profiling and "spiky" LEDs, and better "full spectrum" lights. Then I came across this webinar video (https://dtdch.com/cultural-heritage-webinars/lighting-cultural-heritage/) hosted by Robin Myers and Doug Peterson that seemed to explain this topic very well.
Full spectrum high CRI light bulbs don't seem to be all that expensive, so why are the LEDs in the scanners so uneven? Would I be more likely to archive good reproductions using the camera method where I can control the lighting myself? Just some thoughts I've been having lately.
(Also, how were you able to analyze the light of your scanner? Did you just hold your spectrophotometer in place while the scanner light moved under it? I don't have an android device to use the ArgyllPRO app.)
-
If you are getting the same results with a range of different charts and profiling packages, then that points to either a workflow issue outside the profiling process, or a spectral mismatch between the media you are scanning and the charts. Lacking a spectral characterization of your scanner, the only way of guaranteeing optimal results using test charts and tri-stimulus profiling is to have test charts based on identical paper and colorants to the media you wish to scan.
Is it possible (for me) to create a spectral characterization of the scanner? When I read my targets with Argyll I see that it can include spectral measurements… not sure what to do with that info though. Is there a source for more information about this method? Thanks.
-
Is it possible (for me) to create a spectral characterization of the scanner? When I read my targets with Argyll I see that it can include spectral measurements… not sure what to do with that info though. Is there a source for more information about this method? Thanks.
Yes, but it's not at all easy. One method that has been used with variable success, is to use a test chart (or combination of different test charts) made with a range of independent colorants (so that the spectral characteristics are varied), and then try and figure out what scanner spectral response best explains the measured results. This is a rather under constrained problem, and various tricky mathematical/algorithmic methods are used to try and end up with a sensible result. Typically though the accuracy is moderate at best, and very poor at worst. If you go searching with the right keywords, you will find several dozen academic papers on this type of approach. I don't know of any successful commercial products though.
The other approach is more traditional - spectrally characterize the components of the scanner. Light source is relatively easy, but the sensor is not - the thorough approach is a monochrometer used with a broad band light source to scan through the spectrum and measure the sensor output. In theory the manufacturers of the scanners should have done that, but they don't seem to publish it, not will every example of a particular scanner be the same.
There are other approaches that could be used, but they also rely on some special equipment - i.e. a scan of a linear variable bandpass interference filter. Such filters are not cheap (https://www.edmundoptics.com/p/linear-variable-vis-bandpass-filter/30623/).
-
Is it possible (for me) to create a spectral characterization of the scanner? When I read my targets with Argyll I see that it can include spectral measurements… not sure what to do with that info though. Is there a source for more information about this method? Thanks.
No. Not possible. Further, there is no way to tell what the color filter spectral response used by the scanners is. Perhaps they have been modified to sort of invert the LED spectral bumps and produce something more D50 like: However, I doubt it.
My results are somewhat similar. Interestingly, using the scanner profile I generated from a large printed target produces the almost the same average dE00s as that that built in to the V850 scanner and the custom profile generated from the IT8 target with I1PRofiler.
There is a slight improvement when the reflectance glare is removed.
XRite "Classic" Colorchecker scanned v I1Pro2 measured
dE00: 2.91, Epson Standard, ICC mode
dE00: 3.00, Scanned using I1P IT8 profile, ICC mode
dE00: 3.01, Scanned using 1914 printer patch profile
dE00: 2.74, Scanned using 1914 printer patch profile, refl corrected
Note that the reflectance factor is quite small compared to my tests where patches are either surrounded by black or white.
Also, I believe the good results scanning the same IT8 image used to create a profile overstates the actual, even when considering that the IT8 is unchanged spectrally. The effects of reflected light will be the same. The way to truly test this is to do a scan with randomly re-arranged IY8 patches. I would expect them to have significantly higher errors. I see this using printed targets where I can easily print the same patches re-arranged.
-
I have completed a substantial amount of research about exactly this set of issues and the article is complete, awaiting final comment from several peer reviewers before publication on this website. I expect you'll see it within the next two weeks. So I won't say much here just now, except to note that I too have round-tripped the scanner profiling targets as very much like Brian did as PART of the verification routine. While the profile is built from these patches, that is not the main reason for using them in a round-trip test. There are three other reasons: firstly, we have the file reference numbers for those patches from which to calculate dE values, secondly they are most likely to be within the gamut of the profiles they help to create and thirdly these reference values are also used by the profiling software vendors' own internal validation routines, allowing us to compare outcomes between their tests and ours - to some extent.
I think the results Brian got in the CTP round-trips using the profiling targets are very good. I also think like him that the other results he got leave a fair bit to be desired, so the question is why. There could be a number of reasons. One of the more important ones may be that a large difference between the profiling target media and the profile testing media could create differences of reflectance that create larger dE issues. Another could be the fidelity of the reference values for the test media.
It is also necessary to break the analysis and the discussion into at least two components: one being reflective targets and the other transmissive. While the principles of profile verification are essentially the same, there are differences in the details of application largely having to do with reference values and how they are derived.
Finally, I have observed that scanner profiling is more finicky than printer profiling. More can go wrong and the outcomes tend to show less accuracy. The real issues are the sizes of the inaccuracies, how easily they can be compensated in a post scan editing workflow and whether they really matter to the acceptability of the final results. The smoothness of tonal transitions is something I pay a fair bit of attention to, because while several dE of colour inaccuracy may not amount to a big deal, jagged movements up a tone curve could be an altogether different matter, depending on their size and shape.
Stay tuned.
-
Yes, but it's not at all easy. One method that has been used with variable success, is to use a test chart (or combination of different test charts) made with a range of independent colorants (so that the spectral characteristics are varied), and then try and figure out what scanner spectral response best explains the measured results. This is a rather under constrained problem, and various tricky mathematical/algorithmic methods are used to try and end up with a sensible result. Typically though the accuracy is moderate at best, and very poor at worst. If you go searching with the right keywords, you will find several dozen academic papers on this type of approach. I don't know of any successful commercial products though.
The other approach is more traditional - spectrally characterize the components of the scanner. Light source is relatively easy, but the sensor is not - the thorough approach is a monochrometer used with a broad band light source to scan through the spectrum and measure the sensor output. In theory the manufacturers of the scanners should have done that, but they don't seem to publish it, not will every example of a particular scanner be the same.
Taking apart a scanner and measuring it's spectral response would seem to be more straightforward and more accurate. I'm surprised there are as many studies on this (various spectrally reflectance colors) as you indicate Graeme. Has their been a change since "white" LEDs became ubiquitous? And any idea how that has effected what color filters are used in scanners for their RGB response?
And what about the issue of large area flare crosstalk? This, at least, is something that can be corrected from the scanned RGB values unlike spectral metameric failure.
There are other approaches that could be used, but they also rely on some special equipment - i.e. a scan of a linear variable bandpass interference filter. Such filters are not cheap (https://www.edmundoptics.com/p/linear-variable-vis-bandpass-filter/30623/).
-
I have completed a substantial amount of research about exactly this set of issues and the article is complete, awaiting final comment from several peer reviewers before publication on this website. I expect you'll see it within the next two weeks. So I won't say much here just now, except to note that I too have round-tripped the scanner profiling targets as very much like Brian did as PART of the verification routine. While the profile is built from these patches, that is not the main reason for using them in a round-trip test. There are three other reasons: firstly, we have the file reference numbers for those patches from which to calculate dE values, secondly they are most likely to be within the gamut of the profiles they help to create and thirdly these reference values are also used by the profiling software vendors' own internal validation routines, allowing us to compare outcomes between their tests and ours - to some extent.
I think the results Brian got in the CTP round-trips using the profiling targets are very good. I also think like him that the other results he got leave a fair bit to be desired, so the question is why. There could be a number of reasons. One of the more important ones may be that a large difference between the profiling target media and the profile testing media could create differences of reflectance that create larger dE issues. Another could be the fidelity of the reference values for the test media.
It is also necessary to break the analysis and the discussion into at least two components: one being reflective targets and the other transmissive. While the principles of profile verification are essentially the same, there are differences in the details of application largely having to do with reference values and how they are derived.
Finally, I have observed that scanner profiling is more finicky than printer profiling. More can go wrong and the outcomes tend to show less accuracy. The real issues are the sizes of the inaccuracies, how easily they can be compensated in a post scan editing workflow and whether they really matter to the acceptability of the final results. The smoothness of tonal transitions is something I pay a fair bit of attention to, because while several dE of colour inaccuracy may not amount to a big deal, jagged movements up a tone curve could be an altogether different matter, depending on their size and shape.
Stay tuned.
Yup. That sort of thing is far worse than larger dE errors that occur smoothly. The reflected light errors do not produce abrupt, jagged color changes and are much less objectionable, if even visually noticeable, than things like color shift on darker neutrals compared to lighter neutrals.
The larger errors associated with reflected light errors (wide area crosstalk) materially impact the use of a scanner to calibrate printers or measure density but have much less effect on the visual appearance of scans. Reflected light affects widely separated colors much more than adjacent color patches and it's much harder to see dE differences when widely separated. And it has no impact on the smoothness of color transitions.
-
I'm surprised there are as many studies on this (various spectrally reflectance colors) as you indicate Graeme.
Most such papers relate to spectrally characterizing cameras using test charts. Naturally this is applicable to scanners too (assuming that it's possible to measure the scanner light source.)
-
BrianToth,
I just ran similar experiments on my CCSG. Measured with an I1Pro2.
The V850 had a very similar shift in the neutrals with the b* averaging about -5. using the IT8 chart profiled by I1Profiler.
If I used the profile generated by Argyll from my printed patch set the average of the neutrals decreased a bit but was still blue shifted with the b* running about -4.
Interesting in that scanning neutral colors on printed patches produces very good results with a dE less than 1 on both the colored and neutral printed patches.
It's pretty clear that the V850 LED and whatever color filters they use are a long, long way from L/I.
Also, and rather curiously, the printed patch scanner profile is better than the IT8 scanner profile when processing a scanned CCSG though both show significant metameric error.
Presumably the IT8 profile would work best when scanning things like color prints on Kodak paper, specifically the Endura paper the IT8 uses. But I have no way to verify that since one would need an independent color patch set different from the IT8 but on the same paper.
On the positive side, I can scan my prints, fixed for reflection, and get very good results when using my printed patch set and Argyll. Colors match with an average dE00 of well under 1. But if I scan anything else, things are off. Especially neutrals which is interesting since actual CC neutrals are quite spectrally smooth, So the error is in the printer's neutrals which are just lumpy enough when combined with the really lumpy scanner LED to produce a material shift. Interesting the same, slightly worse, shift occurs with the IT8 profiles.
-
I ran a comparison of 3 scanner profiles against a measured, printed set of patches completely distinct from the patches used to create the profiles. The profiles were an IT8 I1Profiler profile, and Argyll 2871 patch set profiles The test patches measured with the three profiles were a set of 957 colors distributed evenly across the printer's gamut. The IT8 based profile using I1Profiler was the first test and showed an average dE00 of 2, which is not too bad. Especially considering the spectral characteristics of the printer and IT8 chart are fairly different. Also it's only a few hundred patches and read with a different spectrophotometer than the printed patches.
IT8 chart supplied with V850 and used with I1Profiler to create a profile:
mean dE00: 2.02, mean dE00 best 90%: 1.79 mean dE00 worst 10%: 4.09 Max: 6.22
The expanded large, Argyll patch set performed quite well, dropping the average dE00 below 1.
mean dE00: 0.83, mean dE00 best 90%: 0.72 mean dE00 worst 10%: 1.83 Max: 3.69
Modeling and removing the internal scanner reflection crosstalk with scannerreflfix.exe improved things further.
mean dE00: 0.47, mean dE00 best 90%: 0.39 mean dE00 worst 10%: 1.14 Max: 3.33
The errors associated with scanner internal reflections here are much lower than in the specialized tests where a color was surrounded by a larger area of white or black. That can introduce errors of up to about 6 L*. on near whites.
Reflected light errors, on average, are normally low since most patches have a reflectivity of only about 20% and error is proportional to the % of light reflected and most things are averaged by the re-reflected light from surrounding patches. Errors associated with the additional uncertainty from re-reflected light can be reduced somewhat by using a large number of patches. The bumpiness gets better averaged.
Re-reflected light error can not be corrected using a profile and must be removed with a separate process that estimates the re-reflected light's contributions and removes them.
Update: Ran the same three measurements using Baryta luster paper with very little OBAs. The above data was from Costco Glossy which has very large amounts of OBAs with b* running almost -14 when read with M1. However, all my work, and the above measurements, is done with M2 which eliminates the effect of uV and results in very close agreement with non OBA containing media.
mean dE00: 2.32, mean dE00 best 90%: 2.09 mean dE00 worst 10%: 4.36 Max: 6.35 - using IT8 I1Profiler Profile
mean dE00: 0.74, mean dE00 best 90%: 0.64 mean dE00 worst 10%: 1.68 Max: 3.66 - using 2871 patch printed chart to profile scanner
mean dE00: 0.56, mean dE00 best 90%: 0.48 mean dE00 worst 10%: 1.29 Max: 2.91 - as above but corrected for re-reflection
Conclusion: The V850 scanner illuminant does not contain uV. Consequently, all measurements or self constructed charts such as printed charts used to create scanner profiles should either have no OBAs or be scanned by a spectro using M2 (uV cut).