Why does Kodakchrome receive special treatment – Summary

[guyburns]

I'd like to make sure I have a reasonable understanding of what I have learnt from responses to my previous post (http://www.luminous-landscape.com/forum/index.php?topic=50262), by condensing into a few statements all the information from respondents. Correct me if I'm wrong:

1. The best single reference for understanding colour is Hunt's Reproduction of Color, 6th Edition. I hope it's the best, because I've just ordered a copy.

2. Kodachrome has an inherent blue shift caused by the red-sensing layer (and to a lesser extent, the green-sensing layer) having higher densities at any exposure than the blue-sensing layer (p3, http://www.kodak.com/global/en/professional/support/techPubs/e55/e55.pdf).

3. To ensure neutrals remain neutral when projected, Kodachrome was purposely designed to have a blue shift so as to counteract the yellowish light of projection lamps (Hunt: 44, 229-30).

4. For reasons not definitely known, most other slide films do not have a built-in blue shift.

5. A slide scanner can be calibrated with an IT8 target (to counter any scanning inaccuracies) using software such as VueScan or SilverKeeper which generate a colour profile that can be applied to a raw scan to correct for inaccuracies in the scanner. Free software to do the same thing include Scarce (http://www.scarse.org/), Argyll CMS (http://www.argyllcms.com/), and Lprof (http://lprof.sourceforge.net/help/lprof-help.html).

6. Each IT8 target also needs a IT8/CGATS file that contains the colormetric measurements for that target. This text file is required as part of the calibration process. The CGATS file for Kodachrome can be downloaded from FTP.Kodak.com/GASTDS/Q60DATA and is contained in a folder called K3-Data. Info about the colour coding of the IT8 targets is in the document TECHINFO.pdf.

7. Calibrating a scanner using an IT8 target cannot remove the inherent blue cast in Kodachrome slides, it can only remove inaccuracies in the scanning process.

8. The blue cast can be reduced by applying Levels correction in Photoshop using the gamma adjustment (middle slider). Approximate settings are Red 1.19 and Green 1.04 (see reply 21, http://www.luminous-landscape.com/forum/index.php?topic=50262.msg415697#msg415697)

9. Not only Kodachrome, but all slides intended for projection in a darkened room are designed to have contrast substantially higher than that necessary for optimum reproduction as prints. When slides are projected in darkness, the image appears lighter, and this effect is stronger for darker areas of the image. The result is a loss in perceived contrast, which can be counteracted if the contrast of the slide in increased. To quote from Fairchild (p 10, http://www.cis.rit.edu/fairchild/PDFs/PAP02.pdf):

… when scenes are reproduced on transparency film, to be projected in a darkened room, the physically-measured contrast, expressed in logarithmic coordinates, must be about 1.5 times higher than the original scene in order to create an optimum reproduction. However, printed images viewed in illuminated surroundings are optimal when their physically-measured contrast is equal to that in the original scene. This result is described in detail by Hunt.

10. The increased contrast of a slide (as compared to a print) can be counteracted by the appropriate selection of gamma in Photoshop Levels.

11. CIECAM02 (the most recent CIE colour-appearance model) attempts to simulate the appearance of an image in a different environment (such as a darkened room) by including (among other parameters) luminance information about the background. For example, a scanned slide can be given a similar appearance on screen as it would when being projected in a darkened room. A Photoshop plugin, and sample images, can be downloaded from http://sites.google.com/site/clifframes/ciecam02plugin.

I can find no reason among the replies to my original post why Kodachrome should be more difficult to scan than other films, given a good-quality dedicated slide scanner such as a Coolscan 5000. I would have thought the alleged difficulties could be overcome by ensuring that:

• The scanner is calibrated using an IT8 target and the resulting profile applied to the scanned image. This should remove any colour inaccuracies;
• The inherent blue-cast is reduced in Photoshop Levels by a suitable input of gamma to the red and green channels; and
• the high contrast is corrected by altering the overall gamma, again using Levels.

Kodachrome may indeed be more difficult to scan, but why?


Additional Comments Based on Replies
23 Jan 2011

12. The gamma adjustments mentioned in point 8 will vary depending on the color space of the image. The stated numbers (red 1.19 and green 1.04) should work while the scan in still in a linear scanner profile space. If the scan is in another space, such as sRGB, most likely the gamma adjustments will be different.
See http://www.luminous-landscape.com/forum/index.php?topic=50387.msg416119#msg416119

13. When using scanning software provided by the manufacturer of the scanner (e.g. NikonScan for Coolscan devices), and the scanner has not be calibrated with an IT8 target, selecting the Kodachrome setting will likely give better results than a Kodachrome setting in third-party software. This is because the manufacturer of the scanner has probably made an effort to include the RGB sensor characteristics in the Kodachrome setting, whereas third-party software is unlikely to have access to these characteristics, and will assume a generic characteristic.

See http://www.luminous-landscape.com/forum/index.php?topic=50387.msg416244#msg416244: If Silverfast provided Kodachrome LUTs in its software to substitute a Kodachrome target profile then I hope it is adapted per individual scanner model (lamp spectrum, sensor RGB dyes). If Hamrick had to do the same for Vuescan it must have been a hell of a job for him.

See http://www.luminous-landscape.com/forum/index.php?topic=50387.msg416441#msg416441: … here is what Silverfast says on their own web site about their Kodachrome profiles: “… we have implemented generic Kodachrome ICC-profiles for many supported film scanners …”.  (emphasis added)

[crames]

I think it's an excellent summary.

Regarding item #8, the gamma adjustments required will vary depending on the color space the image is in. The numbers red 1.19 and green 1.04 should work while the scan in still in a linear scanner profile space. If the scan is in another space, such as sRGB, most likely the gamma adjustments will be different. It is easy to do by eye. If the red, green and blue gammas are adjusted so that a neutral mid-tone has equal RGB numbers, that should provide a good starting point. One can also change (reduce) the blue gamma instead of the green.

In answer to your final question, "Kodachrome may indeed be more difficult to scan, but why?", I think the reasons are 1. the high dye density can exceed the dmax of most scanners, and 2. not all scanner software has a magic Kodachrome setting which automatically takes care of items #8 and #9. It's not obvious that the manual adjustments to use are gamma adjustments (which effectively tilt the characteristic curves). Trying to eliminate the blue cast using scanner exposure settings (shifting the curves up or down without tilting) or with white-point/black-point adjustments, just doesn't work very well.

[Mark D Segal]

The Kodachrome problem and a solution for it are well-explained here: http://www.silverfast.com/highlights/kodachrome/en.html. In a nutshell, the most efficient approach is to profile the scanner for Kodachrome using a Kodachrome target and associated reference file. If you intend to go this route, don't wait years - Kodachrome targets will become scarce because the film and its processing laboratories are all off the market - gone. Once the various providers' current inventories of Kodachrome targets are sold, game over for that approach unless you can borrow or buy existing ones from their owners.

[crames]

Yes, buy a Kodachrome target before it's too late.

But Mark, where is there a good explanation of the Kodachrome problem at the link you provided?

Profiling the scanner with a Kodachrome target alone is not enough. Even with a Kodachrome profile, you still have to activate the special Kodachrome mode in Silverfast or other scanning software. If the scanning software doesn't have a Kodachrome setting, or you want to try doing it better yourself, you have to make manual corrections.

[Mark D Segal]

Yes, buy a Kodachrome target before it's too late.

But Mark, where is there a good explanation of the Kodachrome problem at the link you provided?

Profiling the scanner with a Kodachrome target alone is not enough. Even with a Kodachrome profile, you still have to activate the special Kodachrome mode in Silverfast or other scanning software. If the scanning software doesn't have a Kodachrome setting, or you want to try doing it better yourself, you have to make manual corrections.

It's not explained in gory technical detail there, but they do say in a nutshell what the basic problem is. And yes, scanning software should have a Kodachrome setting, which most versions of SilverFast and Vuescan both provide. Using that setting with a scanner profile made from a Kodachrome target should provide very decent colour reproduction. Of course the whole idea of an efficient workflow is to minimize the amount of manual intervention needed.

[dmerger]

It is certainly worth a try to profile your scanner for Kodachrome. If you don’t have profiling software, you may want to download one or more of the free profiling software packages available via the internet.  I don’t have the links handy, but should be easy to find with a search.  

I’ve used it to profile my scanner for Fuji Velvia and Provia.  The profiles produced a very slightly different look than the generic profile I was using and other film specific profiles made for my scanner model.  They all produced very similar looks, however, and I can’t say which the best was.  I’ve read that my experience is not unusual.  Profiling may help some, but it may not provide as big a benefit as you might hope.

Maybe you’ve already tried using the white balance tool in Camera Raw or Lightroom.  If not, you may want to give it a try.  I’ve found that tool to be the most effective and easiest tool to use tool to remove color casts.  In some cases white balance can’t fully remove a color cast, so after using white balance you may also need to use the split toning tool or some other method in Camera Raw, Lightroom or PS.  I’ve had the most success with such methods.  YMMV

If you haven’t already done so, you may want to check out the “Kodachrome Profiles” article and the “Scanning Guide” available here: http://www.hutchcolor.com/CMS_notes.html

EDIT: Sorry, I just noticed that you are already familiar with the free profiling software.

Profiling the scanner with a Kodachrome target alone is not enough. Even with a Kodachrome profile, you still have to activate the special Kodachrome mode in Silverfast or other scanning software. If the scanning software doesn't have a Kodachrome setting, or you want to try doing it better yourself, you have to make manual corrections.

For a different perspective, you may want to note this quote from the scanning guide: “All traditional scanner adjustments such as highlight and shadow, cast removal, color correction, sharpening, etc. should be done in Photoshop 7 or later.”  Of course, you’ll have to decide whether this advice from Don Hutcheson is best for your situation.

[Mark D Segal]

Yes, Don's stuff is well-researched and great, but get ready to roll-up your sleeves - only a 32 step process to make a good Kodachrome profile without Kodachrome, and best to have a high-end display and a light box along for the ride. Two of the things SilverFast users appreciate are the IT8 profiling process and the grey balance tool (they call it a "pipette" - OK). I've done and published a fair bit of experimentation with their IT8 process, and on the whole I think it's pretty good. They've also had it independently tested by a lab in Germany. It couldn't be easier to profile a scanner - basically a one button-push operation once you have their target. One can always not bother to profile the scanner, but it does save a lot of work especially with Kodachrome, to use a good profile.

If you still need it, the grey pipette, although not the most user-friendly little gizmo to isolate on their GUI - once you get used to that - produces very good colour balance with just about a click or two. The key to success with the pipette is to isolate a few pixels which really should be GRAY. The tool gives you an iterative process of up to four samples. I've found it very effective. This isn't to open a new debate about whether to adjust images pre or post-scanning, but just to mention that these capabilities are there in various versions of the scanning application and they can do a pretty good job, saving work later on.

[Ernst Dinkla]

Some extra comments on Kodachrome scanning: Kodachrome target versions, Kodachrome versions.

http://members.klosterneuburg.net/handerle/kodachrome.html

If Silverfast provided Kodachrome LUTs in its software to substitute a Kodachrome target profile then I hope it is adapted per individual scanner model (lamp spectrum, sensor RGB dyes). If Hamrick had to do the same for Vuescan it must have been a hell of a job for him.


Remove line 3 and 4 in the summary I suggest, it confuses. Or just mention a blue cast in scanning before white/grey/black is fixed. Any slide film CMY dye set has to be carefully selected to get an acceptable color in projection light. Two daylight slide films should deliver a similar/acceptable image in projection light, that is the goal. The differences can be more saturated etc to please the crowd that buys the film but in the end your greys should be neutral in projection light, the colors represent the scene more or less and the gamut a bit balanced. I have not seen Hunt's comments how Ektachrome dyes were selected, layers tweaked to get there but similar obstacles must have been along his path then.

The moment you change projection light, change the projection screen, let the cat watch the images ...... colors (typical Ektachrome, Fujichrome, Ektachrome) are no longer similar/matching/acceptable. Observer is changed, illumination is changed. If that delivers a significant blue or red or green cast in one of them to the observer then his conclusion should be color inconstancy due to illumination and observer and not describe it as typical xxxx color. The film was created to be exposed in daylight at the right speed, exposure time, normally developed and projected with the standard projector lamp, heat shield, optics, projection screen, average human eye.

With 90% of color scanning dealing with dyes related to Ektachrome films you can expect that the hardware-software is more directed to that task.


met vriendelijke groeten, Ernst Dinkla

Try: http://groups.yahoo.com/group/Wide_Inkjet_Printers/

[Mark D Segal]

Some extra comments on Kodachrome scanning: Kodachrome target versions, Kodachrome versions.

http://members.klosterneuburg.net/handerle/kodachrome.html

If Silverfast provided Kodachrome LUTs in its software to substitute a Kodachrome target profile then I hope it is adapted per individual scanner model (lamp spectrum, sensor RGB dyes).
met vriendelijke groeten, Ernst Dinkla

Try: http://groups.yahoo.com/group/Wide_Inkjet_Printers/

The way it works in SilverFast is that you select the media in the General Tab. If you select Kodachrome, the program defaults to the canned profile LSI provides for Kodachrome AND THAT SCANNER. LSI's canned profiles are made PER SCANNER MODEL. How well their canned profiles perform in your scanner depends on how closely your scanner's behaviour replicates the one they used for generating the profile, so the better behaved the scanner (in terms of performance consistency from one unit to the next) the more satisfactory the canned profile and the less the need for a custom profile. (If you've made a good custom profile, best to use it.)

Of course one other hooker in all of this which seldom gets discussed, but is perhaps the stickiest wicket, is the fact that over many years these transparencies experience colour shifts. The character and extent of it varies depending on a number of storage factors. I've scanned some of my Kodachromes now more than 50 years old - some required a fair bit of "restoration and rebalancing", others much less. No profile can deal with that issue. There are software fixes for this phenomenon but I've generally found manual correction more satisfactory in these cases. Often finding a good black point and a good gray point is all that's required, perhaps aided with a bit of a saturation and contrast boost.

[guyburns]

Re Crames' comment:

Profiling the scanner with a Kodachrome target alone is not enough. Even with a Kodachrome profile, you still have to activate the special Kodachrome mode in Silverfast or other scanning software. If the scanning software doesn't have a Kodachrome setting, or you want to try doing it better yourself, you have to make manual corrections.

I didn't mention the Kodachrome setting in my summary because I wasn't sure if my understanding was correct. So I have another question based on the two settings available in NikonScan (POS or Kodachrome). Does it matter which slide setting you use, as long as you use the same setting when scanning as you did when calibrating? i.e you can scan Kodachrome under a POS setting as long as you calibrated the Kodachrome target using the POS setting. For the case of this hypothetical, assume that ICE is not used – scanning Kodachrome with ICE under a POS setting causes problems.

[Ernst Dinkla]

The way it works in SilverFast is that you select the media in the General Tab. If you select Kodachrome, the program defaults to the canned profile LSI provides for Kodachrome AND THAT SCANNER. LSI's canned profiles are made PER SCANNER MODEL. How well their canned profiles perform in your scanner depends on how closely your scanner's behaviour replicates the one they used for generating the profile, so the better behaved the scanner (in terms of performance consistency from one unit to the next) the more satisfactory the canned profile and the less the need for a custom profile. (If you've made a good custom profile, best to use it.)

Of course one other hooker in all of this which seldom gets discussed, but is perhaps the stickiest wicket, is the fact that over many years these transparencies experience colour shifts. The character and extent of it varies depending on a number of storage factors. I've scanned some of my Kodachromes now more than 50 years old - some required a fair bit of "restoration and rebalancing", others much less. No profile can deal with that issue. There are software fixes for this phenomenon but I've generally found manual correction more satisfactory in these cases. Often finding a good black point and a good gray point is all that's required, perhaps aided with a bit of a saturation and contrast boost.

I did understand from another article that the Silverfast LUTs were not just a replica of the Kodachrome target profile corrections but more adapted to get faster to a good scan result.

On fading I have made my comments already in the other thread. Dark fading must have had little influence so far but projected Kodachrome slides suffered and then with magenta loss. But there were also differences between the Kodachrome 25 <> Kodachrome 64/200 at the time they were all fresh, which goes back to 2001.


met vriendelijke groeten, Ernst Dinkla

New: Spectral plots of +220 inkjet papers:
http://www.pigment-print.com/spectralplots/spectrumviz_1.htm

[Mark D Segal]

I did understand from another article that the Silverfast LUTs were not just a replica of the Kodachrome target profile corrections but more adapted to get faster to a good scan result.

On fading I have made my comments already in the other thread. Dark fading must have had little influence so far but projected Kodachrome slides suffered and then with magenta loss. But there were also differences between the Kodachrome 25 <> Kodachrome 64/200 at the time they were all fresh, which goes back to 2001.


met vriendelijke groeten, Ernst Dinkla

New: Spectral plots of +220 inkjet papers:
http://www.pigment-print.com/spectralplots/spectrumviz_1.htm

Hi Ernst, yes indeed you did comment on the fading issue in the other thread. Asleep at the switch, I hadn't seen that thread until after I posted in this one.

More generally on the Kodachrome issue, Kodachrome does not have an in-built colour cast for dealing with projection factors. The bluish rendition is the result of how CCDs read the spectrum of Kodachrome dyes. SilverFast deals with this in two ways: partly from the Kodachrome profiles it provides, and partly with other parameter adjustments which are triggered by selecting the "Kodachrome" media in the General tab. Both the profiles and these parameter adjustments are scanner-model specific in SilverFast; but those who make their own scanner profiles will have scanner-specific adjustments at least for the profiling part of the adjustment process. In the specific case of SilverFast IT8 targets and profiles, these are for the DIN ISO 64 version of the film. Presumably that would make them less accurate for scanning ASA 25 Kodachrome (and reaching further back in history ASA 10), but most likely still far better than nothing. Readers may also refer to Figure 14 and surrounding discussion of the benefits of SilverFast's IT8 profiling for Kodachrome in my review of the Plustek 7600iAI film scanner on this website, where I report on both qualities and issues, depending on the scanner model and whether the profile is *canned* or *custom*.

By the way - a bit off-topic, this article http://en.wikipedia.org/wiki/Kodachrome, seems to be very up-to-date insofar as it reports that although the processing of Kodachrome was slated to disappear in December 2010, now in January 2011 Dwaynes is processing a large batch of Kodachrome it received last December.

[dmerger]

LSI's canned profiles are made PER SCANNER MODEL.

Both the profiles and these parameter adjustments are scanner-model specific in SilverFast;

Mark, how do you know?  Can you direct me to the portion of Silverfast’s user’s manual or web page that confirms your statements?

The reason I ask is because the Silverfast profiles I tested clearly were not made for my particular model of scanner and were not scanner specific, nor were the profiles very good.

Also, here is what Silverfast says on their own web site about their Kodachrome profiles: “… we have implemented generic Kodachrome ICC-profiles for many supported film scanners …”.  (emphasis added)

[crames]

...Asleep at the switch, I hadn't seen that thread until after I posted in this one.

More generally on the Kodachrome issue, Kodachrome does not have an in-built colour cast for dealing with projection factors. The bluish rendition is the result of how CCDs read the spectrum of Kodachrome dyes...

If you read it I think you will see that proof was provided in that thread that Kodachrome does indeed have a built in color cast (and increased contrast) for dealing with projection factors.

It's possible that the interaction of the scanner CCD, light source, and film dyes will produce an additional blue tendency, but such defects in rendition caused by the hardware are removed (or greatly minimized) by a good scanner profile. The point is that, even after a Kodachrome profile is applied, there is an inherent blue bias due to the characteristic curves of the film itself, that remains to be dealt with by means other than the profile.

It's convenient that Silverfast and other software has a special setting to resolve the problem, but the OP's original question was, why is the special setting even needed?

[Mark D Segal]

Mark, how do you know?  Can you direct me to the portion of Silverfast’s user’s manual or web page that confirms your statements?

The reason I ask is because the Silverfast profiles I tested clearly were not made for my particular model of scanner and were not scanner specific, nor were the profiles very good.

Also, here is what Silverfast says on their own web site about their Kodachrome profiles: “… we have implemented generic Kodachrome ICC-profiles for many supported film scanners …”.  (emphasis added)

I wouldn't write this stuff if I didn't know what I'm talking about, be it from research, testing or reliable sources. I don't know what model scanner you have, what version of the program you are using, whether you have all their profiles loaded into your profiles folder where the program can pick them up and whether you have read the profile names of the profiles you tested - pilot error can happen to the best of us. They haven't profiled every scanner on earth, but they've done a large number of them. Clearly if you used a profile not meant for your equipment it may not yield particularly satisfying results. What they call a "generic" profile, means that it is generic to the scanner model in the sense that it isn't a custom profile for the user's particular scanner. I don't see any other way of interpreting that statement in its context.

[Mark D Segal]

If you read it I think you will see that proof was provided in that thread that Kodachrome does indeed have a built in color cast (and increased contrast) for dealing with projection factors.

It's possible that the interaction of the scanner CCD, light source, and film dyes will produce an additional blue tendency, but such defects in rendition caused by the hardware are removed (or greatly minimized) by a good scanner profile. The point is that, even after a Kodachrome profile is applied, there is an inherent blue bias due to the characteristic curves of the film itself, that remains to be dealt with by means other than the profile.

It's convenient that Silverfast and other software has a special setting to resolve the problem, but the OP's original question was, why is the special setting even needed?

The quote from Hunt is a hypothesis not a proof. That said, even if both factors were true, I think we're agreed that scanners do interpret this film with a blue-ish bias which needs a combination of profiling and other tweaks to neurtralize it.

[dmerger]

This thread seemed odd to me, and now I think I may know why.  Silverfast recently launched its promotion of its “improved Kadachrome workflow.”  Now, all of a sudden we have a guy who registers on this web site on January 11, 2011, and the very next day posts a question about scanning Kodachrome.  Coincidence?  Then, of course, we have the usual promoters of Silverfast extolling the wondrous virtues of Silverfast with posts that read like something straight out of a marketing campaign.


Of course, we’ll probably never know for sure if any of these guys are shills, but the coincidence, if that is what it is, is extraordinary.

[Mark D Segal]

The quote from Hunt is a hypothesis not a proof. That said, even if both factors were true, I think we're agreed that scanners do interpret this film with a blue-ish bias which needs a combination of profiling and other tweaks to neurtralize it.

Crames, an addendum on this issue, as it has further tweaked my curiosity. I went back to a bunch of Kodachromes from 1959, which I am pleased to observe are remarkably well preserved in terms of colour. So 52 years on....... not bad. Fortunately, we are having a neutral gray day here in Toronto and it is just after noon, so the colour temperature of the light would be about as neutral daylight as mother nature provides. And of course, inside the house I have the usual halogen, incandescent etc. lighting. So I picked-up a number of these images with subject matter making it somewhat easy to detect colour casts. Admittedly, this is visual stuff - no instruments, and not always easy judgments to make; on several of them, I didn't see a blue cast under daylight for which the colours became a whole lot better examined under warmer light, but on some of them I did. So Hunt's hypothesis may have something to it, regardless of other information I've accessed. That said, there's no question the scanner itself plays a major role in how Kodachrome is interpreted - I've determined, for example, that the Epson V750 (CCFL) profiles Kodachrome with better neutrality than I could achieve from the flagship Nikon SC500ED (LED), both using SilverFast's Auto IT8 system and the same target.

[crames]

The quote from Hunt is a hypothesis not a proof. That said, even if both factors were true, I think we're agreed that scanners do interpret this film with a blue-ish bias which needs a combination of profiling and other tweaks to neurtralize it.

Are you referring to the quotes at the bottom of this post? http://www.luminous-landscape.com/forum/index.php?topic=50262.msg415789#msg415789

I don't see that so much as a hypothesis, but more a statement of fact by someone who spent 36 years in the Kodak Research Laboratories, is a pioneer in Color Appearance Models, and one of the more eminent color scientists in the world!

Anyway, if the statement by a Kodak color scientist is not enough, and characteristic curves themselves are not enough proof, I offered a demonstration here http://www.luminous-landscape.com/forum/index.php?topic=50262.msg415697#msg415697 that showed that editing a scan with the same gamma corrections that would serve to align the characteristic curves has the predicted effect of eliminating the blue cast that remains after the application of a Kodachrome profile. 

I see in your addendum message below that maybe you are softening your position on this a little, so thanks for taking another look at it.

Cliff

[Mark D Segal]

Are you referring to the quotes at the bottom of this post? http://www.luminous-landscape.com/forum/index.php?topic=50262.msg415789#msg415789

I don't see that so much as a hypothesis, but more a statement of fact by someone who spent 36 years in the Kodak Research Laboratories, is a pioneer in Color Appearance Models, and one of the more eminent color scientists in the world!

Anyway, if the statement by a Kodak color scientist is not enough, and characteristic curves themselves are not enough proof, I offered a demonstration here http://www.luminous-landscape.com/forum/index.php?topic=50262.msg415697#msg415697 that showed that editing a scan with the same gamma corrections that would serve to align the characteristic curves has the predicted effect of eliminating the blue cast that remains after the application of a Kodachrome profile. 

I see in your addendum message below that maybe you are softening your position on this a little, so thanks for taking another look at it.

Cliff

Of course, one always wants to take another look when presented with credible alternative information. Given the history of Dr. Hunt's Kodak research affiliation you mention and the evidence shown, I agree with you that we should take those statements as a definitive explanation of Kodachrome's colour reproduction approach and the reason for it. It does make sense. At the same time, still open are other questions about why Kodachrome behaves better in some scanners than in others using the same software and profiling process. So summing up, we have a clear enough picture now of why Kodachrome is different, the fact that it reacts to scanning differently from other transparency materials and in different scanners, and needs both separate profiles and other tweaks in the scan algorithm to obtain neutral neutrals. It's been an informative discussion. Thank you.