Ad
Ad
Ad
Pages: [1]   Bottom of Page
Print
Author Topic: Very Unexpected Scanner Profiling Results  (Read 2647 times)
dmerger
Sr. Member
****
Offline Offline

Posts: 686


« on: November 01, 2012, 12:19:16 PM »
ReplyReply

I’ve been making new profiles for my Minolta 5400 scanner using CoCa profiling software (it uses ArgyllCMS) and Wolf Faust IT8.7 Fuji Velvia 50 and Provia targets.  I set my Minolta scanner software to output 16 bit linear tiffs without any color profile, which I used to make my profiles.  (In CoCa I used the XYZ algorithm, high quality.)

I’ve created profiles directly from the linear files, and also first made 1.8 gamma adjustments to the files and then created profiles.  Also, CoCa has a tab to select the gamma it uses (i.e. 2.2 for a normal gamma adjusted image, gamma 1 for linear, or gamma 1.8, etc).  I’ve tried several combinations, e.g. a linear file using CoCa’s 2.2 and 1 gammas, and 1.8 gamma adjusted files using CoCa’s 2.2 and 1.8 gammas.  

The unexpected outcome is that all the profiles produce identical results.  Not only do the results visually look identical, but I’ve also compared them by opening two copies of an IT8.7 target photo using different profiles and then converting them to ProPhoto so I could stack them in layers in PS.   I set the top layer to “difference”, and it showed no difference at all.  

I’ve even opened the Provia target in PS, and then assigned one of the Provia profiles I created, and also opened the same target but assigned the Velvia profile, and the two are identical.  In other words, a profile created using the Provia IT8.7 target produces the identical profile created using the Velvia IT8.7 target.

These results are not what I expected.  Are my results normal?  

(The profiles do produce nice looking results.  I did notice, however, that these new profiles look noticeably different than the profiles I made a few years ago using the same IT8.7 targets and other profiling software.  I don’t have any software to show which are most accurate, but visually I like the results from the new profiles better.  They are brighter and seem to more closely match the slides on my light box.  EDIT: After editing a couple photos in LR (not just looking at the IT8.7 targets), it seems like the canned Minolta linear profile produces the better appearance. )

Second EDIT: I did another comparison.  I compared the Provia IT8.7 target after increasing the exposure in PS for the Minolta profiled version to equal the brightness of the CoCa profiled version.  Now, the two are much closer, but the CoCa version is noticeably more blue (cooler) than the Minolta version. I also tried to tell visually which version more closely matched the slide on my light box.  (I use an NEC 271w with SpectraView.)  It’s difficult to be sure, but I’d say that the Minolta version might be a little better match. In other words, both are close, but the Minolta may be a little more accurate, even though the numbers don’t bear that out according to my tests described below.)
« Last Edit: November 01, 2012, 05:18:24 PM by dmerger » Logged

Dean Erger
digitaldog
Sr. Member
****
Offline Offline

Posts: 9225



WWW
« Reply #1 on: November 01, 2012, 01:08:45 PM »
ReplyReply

The unexpected outcome is that all the profiles produce identical results.  Not only do the results visually look identical...

Actually it isn't. The custom profile should account for the differences visually. The question is, which setting provides the best data?
Logged

Andrew Rodney
Author “Color Management for Photographers”
http://digitaldog.net/
dmerger
Sr. Member
****
Offline Offline

Posts: 686


« Reply #2 on: November 01, 2012, 02:06:48 PM »
ReplyReply

The question is, which setting provides the best data?

I agree, but how do I determine which is more accurate?  Is there some way to do so short of buying additional profiling software? 

Of course I’ve got the data files for my IT8.7 targets, but that data does not appear to correspond directly to the RGB or LAB data I see in PS.
Logged

Dean Erger
digitaldog
Sr. Member
****
Offline Offline

Posts: 9225



WWW
« Reply #3 on: November 01, 2012, 02:10:02 PM »
ReplyReply

You could plot the gamut's of all, see if there's a difference.

You could scan and reduce each color in the target to a single color, then create a color list in ColorThink. Now you can produce output values for each scan. What do you want to compare it to? The reference of the target?

You say you like the look and all look about the same. So you could just toss a coin and be done. I suspect one of the settings should produce a slightly better numeric result (once we decide what to use to compare it to). If you have the time and software, it be neat to run the numbers on the various scans.
Logged

Andrew Rodney
Author “Color Management for Photographers”
http://digitaldog.net/
dmerger
Sr. Member
****
Offline Offline

Posts: 686


« Reply #4 on: November 01, 2012, 04:07:22 PM »
ReplyReply

Andrew, I have the time, but not the software nor the knowledge to do what you recommend.  Smiley

I compared the LAB numbers in the IT8.7 reference file to readings in PS.  The IT8.7 numbers are to two decimal places, and PS using the color sampler in 8 bit mode is in whole numbers.  Since the PS numbers are usually off from the IT8.7 numbers by more than 1, sometimes a lot more, however, I think I can safely ignore rounding errors.

I compared my Provia target (1) with the CoCa profile converted to ProPhoto, and (2) with the Minolta profile converted to ProPhoto. I looked at four patches from the grey scale and four from each of the red, blue and green columns (columns 17, 18 and 19).  The “L” numbers in PS are consistently higher than the IT8.7 data, with the CoCa profile consistently higher than the Minolta profile.  The LAB “A” and “B” numbers from the CoCa profile, however, are usually a little closer to the IT8.7 target than the Minolta profile.  So, based on this unscientific test, it appears the CoCa profile may be a little more accurate, but since the CoCa “L” numbers are consistently off by more than the Minolta numbers, maybe not. 

Assuming my little test isn’t totally invalid, with a good profile, how close should the LAB numbers in PS be to those in the IT8.7 reference file?

From a practical standpoint, however, I think you’re correct, Andrew.  Both profiles still require editing in LR to produce a nice looking photo, and the edits in LR have much greater effect on the final photo than does the relatively small difference in the profiles.  Nevertheless, the differences in the profiles are noticeable, even after processing in LR. 

I went through this whole exercise a few years ago, and came to the conclusion that the canned Minolta profile was at least as good as the other profiles I made, or others made on the same model scanner.  All those profiles, however, were much more similar to one another than the new Coca profiles are to the Minolta profile.  So, while I find this exercise to be interesting, I’ll probably just continue to use the Minolta profile.

(Of course this whole discussion about accuracy doesn’t address why all the CoCa generated profiles are all the same.)

Logged

Dean Erger
mless
Newbie
*
Offline Offline

Posts: 2


« Reply #5 on: November 04, 2012, 08:56:38 PM »
ReplyReply

Hello dmerger, digitaldog,

I'm in the process of creating ICC profiles for my scanner using IT8.7 charts and something is also puzzling me about the results. Maybe you guys can help me understand this better.

My problem is the opposite of yours: the profiles I create are always different, even if I use the exact same IT8 chart, scanned the exact same way, twice in a row. If I make 2 consecutive scans of the IT8 chart without even opening the scanner lid, and then generate 2 profiles with each of theses 2 scans, the RGB curves in the resulting profiles are often very different (ie not the same shape). The whole thing seems pretty random to me. The scanner is new and works fine.

I simply scan an IT8 chart with Vuescan and save the output as an unprocessed TIFF (Vuescan's "raw" mode) and then use Argyll CMS to produce the ICC profile. I mostly use the default Argyll settings just like in the basic tutorial in the doc. The diagnostic image from Argyll shows it seems to be processing the target fine. Now when I use Photoshop to assign the profiles I created this way to images I've scanned the results are mostly good, not _that_ different from profile 1 to profile 2, just a slightly different tone. But I'm very curious about this... I would expect the profiles to be pretty similar if they are created the same way with the same scanner! Why do I get different RGB curves every time?
Logged
darlingm
Sr. Member
****
Offline Offline

Posts: 349


WWW
« Reply #6 on: November 05, 2012, 01:50:04 AM »
ReplyReply

I'm in the process of creating ICC profiles for my scanner using IT8.7 charts and something is also puzzling me about the results. Maybe you guys can help me understand this better.

My problem is the opposite of yours: the profiles I create are always different, even if I use the exact same IT8 chart, scanned the exact same way, twice in a row. If I make 2 consecutive scans of the IT8 chart without even opening the scanner lid, and then generate 2 profiles with each of theses 2 scans, the RGB curves in the resulting profiles are often very different (ie not the same shape). The whole thing seems pretty random to me. The scanner is new and works fine.

I simply scan an IT8 chart with Vuescan and save the output as an unprocessed TIFF (Vuescan's "raw" mode) and then use Argyll CMS to produce the ICC profile. I mostly use the default Argyll settings just like in the basic tutorial in the doc. The diagnostic image from Argyll shows it seems to be processing the target fine. Now when I use Photoshop to assign the profiles I created this way to images I've scanned the results are mostly good, not _that_ different from profile 1 to profile 2, just a slightly different tone. But I'm very curious about this... I would expect the profiles to be pretty similar if they are created the same way with the same scanner! Why do I get different RGB curves every time?

Is your scanner/software creating 8 bit channel depth (also called 24 bit) TIFF files?

Are you using an optical resolution, or an interpolated resolution?  (Most scanners let you scan at a higher dpi than it really sees at, and it resizes the image like Photoshop would.)

If it's creating 8 bit files, you'll see a larger difference than if it's creating 16 bit channel depth (also called 48 bit) TIFF files.  (Note that what matters is the bit depth your scanner "sees" at, and the bit depth of the original file created by your scanner/software.  Some scanners and/or software can only do 8 bit depth, and converting that into 16 bit depth won't make a difference... At least for sending to a profiling program.  If you're doing post editing, you could see a reduction of banding if you move into 16 bit first.)

If you feed the same untagged scanned file into the same profiling software using the same settings, you (darn well should) get identical profiles.  This means that your back to back scaned files are different.

This, unfortunately, is completely expected.  Scanners aren't precise enough.  You can think of this as a compound rounding error.  You're going to have some amount of noise, and a lot more of it than you'd initially think.  The lamp isn't going to emit the same exact light spectrum every time.  Additionally, the temperature of the lamp affects the light being emitted from it.  If you're doing the first scan after it's been idle, the lamp might be warmer the second time through.  Many scanners will "warm" the lamp before a scan after being idle, but there will still be a difference here.  Furthermore, each scanner is going to have its own expected variance from scan to scan, depending on how well it's engineered -- but it could never be perfect.

These variances will cause the same patch to average to different values on the back to back scans.  Since you're only using a few hundred patches to simulate the scanner's response to millions of colors, this relatively small variance is going to make profiles be measurably different, although as you're seeing, not completely different.  And it should.  Profiling software can't tell how accurate a patch is - it just has to assume the values it's given.

These variances will be an even bigger issue at 8 bit depth.  A specific patch might average on the first scan to 200R/80G/60B, but on the second to 201R/81G/61B.  Doesn't seem like a huge difference, but it is noticeable (at least on a calibrated wide gamut monitor) -- and furthermore, this one patch is representing tens of thousands of colors, if we go with the idea that the human eye & brain can distinguish between 10 million colors.  Photoshop may report these same values at 16 bit depth, because instead of showing you either 51364R or 200.64R, it's going to show you 201R.  But, the behind the scenes scale is still going from 0 to 65535 instead of 0 to 255, so there's more precision in the file than Photoshop lets you see in the RGB values.  If 16 bit depth is being used, you're eliminating one step of the rounding errors, so the variance will be less, but there's still going to be noise and variance.  Even though your scanner probably has a few of these bits as random garbage noise, rather than having a truthful 16 bit depth throughout every step along the way, you should still get more than 8 bits of precise data this way.)


You can see this difference between the files in Photoshop.  Since you aren't moving the target or lifting the lid, the target should be positioned at the same point in the image, or at least very close to it.  You can try copying the contents of one file into the other, into a second layer.  Turn this second layer's blending mode into difference.  You will likely see a near black document here.  Create an adjustment curves layer.  Don't create a third point, but drag the top right 255/255 point itself across the left a ways -- somewhere around half way across, or all the way to the left that you can go.  (This curves layer will emphasize subtle differences that would otherwise show as near black and be hard or impossible to notice.)

Idealy, there would be no difference between the scans, however in reality you'll never get identical back to back scans.
Logged

Mike • Westland Printworks
Fine Art Printing • Amazing Artwork Reproduction • Photography
http://www.westlandprintworks.com • (734) 255-9761
digitaldog
Sr. Member
****
Offline Offline

Posts: 9225



WWW
« Reply #7 on: November 05, 2012, 08:56:07 AM »
ReplyReply


My problem is the opposite of yours: the profiles I create are always different, even if I use the exact same IT8 chart, scanned the exact same way, twice in a row.

Sounds like a scanner driver that is doing some kind of auto correction setting each scan. Can't profile a scanner that way. What's the scanner driver and does it actually accept ICC input profiles? If so, you might need to find the right want to configure the driver for the creation of a profile. Some scanners can't be profiled due to this auto behavior (the old Polaroid Sprintscan comes to mind).
Logged

Andrew Rodney
Author “Color Management for Photographers”
http://digitaldog.net/
dmerger
Sr. Member
****
Offline Offline

Posts: 686


« Reply #8 on: November 05, 2012, 04:50:10 PM »
ReplyReply

Mless, do you mean that you scanned your target twice, profiled each scan twice, and all four profiles differ?  If so, then you have a hell of a mess!   Cry 

If you use the same file, profile it twice, and get different results, then you have a problem at the profiling stage.  If the profiles from the same file match, but differ from the file from your other scan, then you have a problem at the scan stage.  If all four profiles differ, then you have a problem at both the scan stage and the profiling stage.
Logged

Dean Erger
mless
Newbie
*
Offline Offline

Posts: 2


« Reply #9 on: November 07, 2012, 04:21:34 PM »
ReplyReply

dmerger: No, if I create a profile twice from the same scan it produces the same results. I use 16 bits per channel images scanned at 600 dpi optical resolution. I'm just surprised that making two scans of the same chart would produce that much difference in the profile curves so I first suspected the profiling software. I understand the 2 scans will always be a little different, but I would expect the differences to be minor and the two scans to produce similar profiles...

digitaldog: The scanner is a basic consumer Canon scanner (LIDE 210) and I installed the driver from the Canon website. I have no idea what the Canon driver does or if they provide a different kind of driver more suited for profiling (I doubt it). I'm surprised that a driver would correct two scans of the same image that differently... So you think trying to profile this kind of scanner is useless?


Thanks!
Logged
Pages: [1]   Top of Page
Print
Jump to:  

Ad
Ad
Ad