Luminous Landscape Forum
Raw & Post Processing, Printing => Colour Management => Topic started by: guyburns on January 12, 2011, 10:59:45 pm
-
I'm about to start scanning my Kodachrome slides, but have been testing how to scan slides in general. What has got me stumped is: why does some software (e.g. NikonScan, Vuescan) have special settings for Kodachrome?
If two people took exactly the same photo with different slide film, and then projected them, you'd expect to see different colours. i.e. Velvia might have a more saturated sky than Kodachrome. The difference is locked into the slide. My thinking is: you've got a bunch of slides and you project them on a certain projector. Whether the slide is Kodachrome, Fuji Velvia, Anscochrome, Olypmus Pen (all of which I'm testing, plus others), you use the one projector and simply accept that different slide stock will have a different look. Why do scanners offer options for various slide film?
There are several possibilities why Kodachrome (or other films) have special settings.
1. The scanning software is trying to see through the slide to what was actually imaged on the day. i.e. if you select Kodachrome in Nikon
Scan, the colours are adjusted in a certain way so that what you see after the scan is what the real scene was like when the slide was taken.
2. The software is trying to adjust for the look of the slide when projected on a slide projector. i.e. to give it a slight yellow cast.
3. A combination of the above?
4. The corrections are just a ploy to make the user think they are getting an improvement. The more options the better, the kind of thinking that comes from marketing people.
I threw in the last comment because on my system, choosing Kodachrome within Vuescan does not alter the colour of the image (as compared to the "generic" setting); whereas NikonScan does visibly alter the colours. The results from 15 test slides, shows that NikonScan (Kodachrome) shifts the colours like this (as compared to the POS setting):
Red increases by between 13 and 25
Green increases by between 5 and 10
Blue decreases by between 5 and 9.
The numbers stated are the mean of the various colour channels as measured in Photoshop using Histogram.
I have not thrown in the complication of faded slides or slides taken with filters (UV, 80A) and so on. Lots of slides have gained a colour cast through age, or acquired one by the addition of a filter. How sensible is it to make colour adjustments during the scan on an already colour-adjusted slide?
I began my testing by thinking that slides should be scanned "as is" and all corrections done later on. Now I'm not so sure. Under NikonScan, for instance, the use of ICE (dust removal), ROC (restoration of colour), GEM (grain removal) and sometimes DEE (highlight/shadow adjustment), gives superb results that cannot be obtained any other way (I've been testing 3 scanners in combination with NikonScan, VueScan and SilverFast). I asked about this in the Photoshop forums (http://forums.adobe.com/thread/766773?tstart=0), but even though 39 people have viewed it and 7 have downloaded the files, no one has replied that PS can do better.
So, what is the idea of a Kodachrome setting?
-
So, what is the idea of a Kodachrome setting?
Kodachrome used a pretty unique dye set to create the colors of the image. That dye set is really kinda weird and Kodachrome is an unusual process in that the conversion of silver to dye was a complex chemical conversion unlike anything else in the industry–ever. The Kodachrome process is incredibly caustic and nasty (one of the reasons Kodak shut down processing because of OSHA standards for dealing with hazardous materials).
This was all started way before the concept of scanning film was ever contemplated...
E-6 film's dye set is pretty easy to scan. Kodachrome is far more difficult to scan in order to get a color response that duplicates the appearance of the original chrome. This is because the dye set to the eye is different than to a scanner (a scanner is not a human eye).
At this point, there's no way to create a scanning target since Kodachrome processing ended forever and the end of Dec, 2010.
About all you can do is to look at the chrome and then adjust the scan to be a close match on-screen to "look" like what you see in the chrome. And that's a twiddly process.
And no, you absolutely need to optimize the tone & color as you scan and not try to fix it after the fact in PS. But, you should be able to tweak settings and be able to save settings for various Kodachrome variants such as old (original) 25, 64 and new 25 and chromes with various states of fade.
Scanning Kodachrome is far more difficult than scanning E-6 films whose dye sets were essentially all the same (meaning the why the dye responded to scanners).
Personally, I always hated scanning Kodachrome because of the work involved getting a really good scan. E-6 is MUCH easier...
-
Thanks for the response, particularly the bit about some corrections should be done while scanning. A lot of comments I have read elsewhere about scanning imply that scanning should be done "clean" and adjustments later. But I'm coming around to the opposite view.
I still don't understand why there are different settings offered for different film stock. Imagine the perfect scanner: it has been calibrated to accurately scan every possible colour. When a slide is put in this scanner, the rich blue of a Velvia sky is rendered perfectly; the more muted blue of the same Kodachrome sky also renders perfectly. I assume that is a correct way of viewing this perfect scanner. Every scan of a slide would appear just like the original slide and Kodachrome would present no problems.
Now down to a real scanner. Assume any colour transfer characteristic you like: the scanner may be in good calibration or even reverse calibration (colours sent to their opposite). When slides are presented to this imperfect scanner, the rich blue Velvia sky may come out as pink; the Kodachrome sky may come out as a more muted pink. The actual colour characteristic are unimportant for this example because the scanner is simply a device that records what it sees, even if it sees colours incorrectly. I can't understand why the slide stock makes a difference when this imperfect scanner is used -- it's just a recorder of colour.
Another thought experiment: imagine some photographic expert taking a Kodachrome slide of a scene using special filters so that the Kodachrome image is identical to an unfiltered Velvia image. i.e. you have an exact equivalent of a Velvia image but it is on Kodachrome. Now scan it. Would this particular Kodachrome slide be more difficult to scan than the identical-looking Velvia slide?
I understand that slides taken on different stock look different, but what is it about Kodachrome colours that make them difficult to scan? Unless you mean that the colour gamut of Kodachrome is such that it contains colours that cannot be accurately scanned. But that can't be the case (can it?) because some Kodachrome colours would then look wrong in a projected slide or any image taken off the slide. If it distorted colours, Kodachrome would not have had the respect that it did.
Kodachrome slides may be difficult to scan, but is the reason because of colour? Could it not be some other characteristic?
-
Kodachrome slides may be difficult to scan, but is the reason because of colour? Could it not be some other characteristic?
Ok, try this; the same color YOU see when you look at a chrome is NOT the same color the scanner sees when scanning the chrome. It's not unlike the way different sensors see the same scene when you use a digital camera.
Kodachrome's dye set is fundamentally different (and more difficult to render) than E-6.
Whether you are scanning Velvia or Ektachrome 64, the dye set (the visible spectral response of the film), is the same. You may need to tweak Velvia differently than Ektachrome 64, but the way the scanner "sees" the color dyes in the film is the same.
With Kodachrome, that's not the case...each version of Kodachrome has different dye sets than any other film. So, scanning Kodachrome is a more unique process than other sorts of film...
It's almost as problematic as scanning color neg film. Color neg film is variable...a certain C-41 line may process a slightly different look months apart. A lab from one area code may not have the same line as a different area code. There is a lot of variation from one C-41 from one line to another.
I would argue that the spectral response of Kodachrome is different enough to be considered an outlier...the "look" between different emulsions was enough to warrant potentially completely different scanners set-ups.
Really, scanning Kodachrome is not "easy".
Once you have a scanner setup for a particular emulsion, it is pretty consistent. The trick is getting to that optimal point with a given type of emulsion...
-
Would it be any easier if the scanner had 20 narrower frequency bands, instead of 3 broad "r", "g" and "b" filters?
-
Would it be any easier if the scanner had 20 narrower frequency bands, instead of 3 broad "r", "g" and "b" filters?
Easier?
Don't know...different prolly...but just know that each Kodachrome emulsion type will have different dye sets which means "different" scanner setting. Can you successfully scan Kodachrome? Yes...but it takes more effort than E-6.
-
Would it be any easier if the scanner had 20 narrower frequency bands, instead of 3 broad "r", "g" and "b" filters?
Multi spectral scanning is used for art reproductions, more colorant types in the originals than the usual CMY dyes of photos. The HP G4050 G4010 consumer scanners do something like that, there are two fluoresent lamps in the scanner and a dual scan is made, one for each lamp. Based on the difference in color reflections between the two scans it should be possible to nail the right color better. One RGB linear CCD though. Image Engineering tested that for reflective photo + print targets and reflective acrylic paint patches. It seems to work better. I doubt that they will improve a Kodachrome film scan, the density range flatbeds like that can cope with is likely lower than the Nikons can. For Kodachrome slides you need that too.
http://www.image-engineering.de/index.php?option=com_content&view=article&id=57&Itemid=91
http://www.lumiere-technology.com/
met vriendelijke groeten, Ernst Dinkla
New: Spectral plots of +220 inkjet papers:
http://www.pigment-print.com/spectralplots/spectrumviz_1.htm
-
There are a couple other issues besides the mentioned interactions between the Kodachrome dyes and the scanner filters and light source.
Because Kodachrome was designed for projection, it has an intentional blue bias to counteract the yellowness of tungsten projector bulbs. If you look at the characteristic curves of KC you can see how the R, G, and B curves are tilted with differing gammas, unlike Ektachrome or Velvia that have the curves aligned for a neutral gray scale.
The other issue is the high overall contrast, which compensates for the low viewing contrast that arises from the dark environment in which slides are projected.
The hardware issues "should" be resolved with a custom ICC profile. Kodachrome IT8 targets are still available at B&H and possibly other sources:
http://www.bhphotovideo.com/c/product/692564-REG/LaserSoft_Imaging_LA1209_35mm_Kodachrome_IT8_Calibration.html
But an ICC profile will not solve the blue bias and high contrast issues. To eliminate the blue bias, the curves have to be tilted to align them - this can be accomplished by adjusting the red and blue channels with gamma adjustments (such as with Levels). Obviously the high contrast can be dealt with in a number of ways, such as with an overall gamma adjustment.
I believe that these are the issues being addressed by the Kodachrome setting of Nikonscan.
Cliff
-
... Could it not be some other characteristic?
When you look at a Kodachrome emulsion at a certain angle, you can practically see a relief on the surface, remotely similar to Find Edges in Photoshop. I would assume this almost 3-D surface presents another challenge for a scanner.
-
I ordered my last roll of Kodachrome from Dwayne's to be processed and scanned -- hope they know how to do it.
-
Ernst: thanks for the various links. I didn't realise the Coolscan 5000 could multiexpose and thus increase the dynamic range. Nice facility to play around with for single scans, but would cause a significant increase in time and probably be not that useful for me given the large number of slides I want to scan. I'm also a bis sus about multiscans because of the chance of additional blurring caused by the extra passes.
Crames: I had wondered myself, but was very doubtful, whether Kodachrome had been pre-adjusted for a blue-cast to compensate for the yellowish light of projection lamps. Do you know if this idea has been documented in Kodak literature or elsewhere?
The idea of high contrast intrigued me, but I think your explanation is not quite right. Dim lighting allows the shadow detail of a high-contrast projected image to be seen on screen. Solid blacks (implying high contrast) are a very important aspect of digital projectors, and one of the main concerns of picky home-theatre enthusiasts (see "What is all this fuss about black levels" under http://www.projectorreviews.com/epson/home-cinema-8700ub/image.php#black). But dense blacks are only possible in dim lighting because the white screen will otherwise wash them away. The way I understand it, high contrast in a slide does not compensate "for the low viewing contrast that arises from the dark environment"; rather, the dark environment allows a high-contrast slide to be viewed in all its shadowy glory.
Colour Casts
This colour cast "thing" has got me baffled. I'm trying to understand where it comes from, but all I get is conflicting information.
Kodachrome has a Red Cast
Kodak's Density Curve (page 3 of http://www.kodak.com/global/en/professional/support/techPubs/e55/e55.pdf, similar to the one above in Crame's post) shows the red curve on top. In an extensive search through Google Books, several authors made the comment that Kodachrome has a red cast. One author, in reference to the Density Curve already mentioned, states:
The characteristic curve shows both a shoulder and a toe. In Kodachrome 64, the red sensitive layer is consistently more sensitive than the green and the blue sensitive layers which accounts for the Kodachrome typical red cast.
I don't yet fully understand these curves, but it seems to me that such authors may have misinterpreted the graph. At a particular exposure, red is on top, but doesn't that mean red is denser, i.e. darker and therefore less visible, throwing a blue-green cast over the image? I'm hoping someone can explain what a Density graph is conveying.
However, maybe the red cast is due to dark fading. In the Focal Encyclopedia of Photography, p 383, it says:
Dyes in colour films and prints can fade under the effect of light as well as in the dark… the terms light fading and dark fading are used to distinguish between them. The dyes present in colour photogrpahs made by the process of chromogenic development are the only dyes know to fade in the dark. There are no other colour documents that behave similarly…
Three variations of chromogenic development processes were developed by manufacturers in the period from the early 1930s to the end of the 1940s. The first commercially available colour slide film was Eastman Kodak's Kodachrome films…
It was observed early that cyan is the least stable dye in the dark… and assumes a reddish overall cast. Under the influence of light, however, cyan dyes are generally the most stable, while magenta is weakest. Photographs faded by exposure to light therefore show a blue-green overall tone, which is caused by the predominance of the surviving cyan plus yellow dyes.
So a red cast could be explained by dark fading (most slides are kept in the dark after initial viewings), but if exposed to light, they can turn blueish. This leads to…
Kodachrome has a Blue Cast
In lots of photographic forums, according to the people who have made comment, Kodachrome supposedly has a blue cast. Maybe it has: in the Kodak PDF already linked to, to the right of the Density graph on page 3 is a graph called Spectral-Dye-Density Curves, shown below. I don't understand the units on the vertical axis: Diffuse spectral density. What it seems to show is that cyan has the highest value. Is this the cause of Kodachrome's blue cast?
I am also confident that the Nikon Coolscans themselves throw a blue cast (of various intensity) on all slides scanned without adjustments. I have come to that conclusion after scanning a wide variety of stock, of varying ages and colour casts, using NikonScan, VueScan and Silverfast. Maybe my scanner is at fault. But if NikonScans throw a blue cast, why would that be?
Kodachrome has a Yellow Cast
Very occasionally this is mentioned by people in various forums. A quote from B&H's website about the IT8 calibration target, explains that CCDs see less yellow than our eyes, and this explains the blue cast seen in scans:
The human eye will see more of the yellow spectrum contained in an Kodachrome image than the scanner's CCD. Consequently scanning a Kodachrome transparency with an ICC-Profile based on a Fuji or Ektachrome IT8-calibration, will produce bluish scans. Only with a dedicated Kodachrome calibration target can correct and original "Kodachrome colors" be achieved.
Of course that's a sales pitch by SilverFast trying to sell a Kodachrome target. Kodachrome is explicity mentioned, but if CCDs really see less yellow than our eyes, doesn't that say that all scanned slides will have a blue cast? Is it documented anywhere that CCDs have a problem with yellow?
Looking forward to any replies, particularly those with links to respected sources. Finally, can anyone explain what this graph is trying to tell me?
-
I have recently finished scanning my father's slide collection. These included slides from the early 1940s and included many shot using 828 film. For the task, I used my Minolta 5400 and Microtek 120tf scanners. I use Silverfast with both scanners and both are calibrated using a IT8 Kodachrome target. The results were more than satisfactory.
BH Photo have the targets in stock - http://www.bhphotovideo.com/c/search?Ntt=koq60k3&N=0&InitialSearch=yes - and they are also available from Lasersoft.
Bob Rapp
-
Crames: I had wondered myself, but was very doubtful, whether Kodachrome had been pre-adjusted for a blue-cast to compensate for the yellowish light of projection lamps. Do you know if this idea has been documented in Kodak literature or elsewhere?
This is from The Reproduction of Colour 6th Edition by R.W.G. Hunt, formerly Assistant Director of Research, Kodak Limited, pg.44:
The reproduction is therefore also required to give a neutral gray impression to the mind, and the eye, seeing a screen lit by tungsten light in a darkened room will be partially, but not quite wholly, adapted to tungsten light. It is therefore necessary that the neutral grey scale be reproduced on the transparency as slightly blue in order to overcome the slightly yellowish appearance of the tungsten light on the screen.
Hunt doesn't mention any specific film, but it is obvious from the characteristic curves that Kodachrome is one of them. I don't know about the older versions, but all the Kodachromes of the last 30-40 years (K25, K64, K200, etc) have the same blue bias in their curves.
The idea of high contrast intrigued me, but I think your explanation is not quite right. Dim lighting allows the shadow detail of a high-contrast projected image to be seen on screen. Solid blacks (implying high contrast) are a very important aspect of digital projectors, and one of the main concerns of picky home-theatre enthusiasts (see "What is all this fuss about black levels" under http://www.projectorreviews.com/epson/home-cinema-8700ub/image.php#black). But dense blacks are only possible in dim lighting because the white screen will otherwise wash them away. The way I understand it, high contrast in a slide does not compensate "for the low viewing contrast that arises from the dark environment"; rather, the dark environment allows a high-contrast slide to be viewed in all its shadowy glory.
No, it's the other way around: the loss of contrast necessitates the higher contrast. The loss of contrast is due to simultaneous contrast effects in the eye and flare effects in the camera and projector. Hunt covers this extensively in his chapter 6 on Tone Reproduction:
If film is projected in a dark room, the surround is normally of a very much lower luminance than that of the picture, and we may call this a 'dark surround'. The effects of these dim and dark surrounds are to make the pictures appear lighter than would be the case with a surround of the same luminance...But this lightening occurs to a greater extent in dark areas than in light areas of the picture...With a dim surround, in order to produce tone reproduction that appears correct, it is necessary to increase the effective objective gamma of the system to about 1.25...with a 'dark surround' the apparent lightening of the dark areas of the picture is even more marked, so that to produce tone reproduction that appears correct now requires a gamma of about 1.5"
You might be interested to read http://www.cis.rit.edu/fairchild/PDFs/PAP02.pdf (http://www.cis.rit.edu/fairchild/PDFs/PAP02.pdf) "Considering the Surround in Device-Independent Color Imaging" by Fairchild, or look up "Bartleson-Breneman effect".
Colour Casts
This colour cast "thing" has got me baffled. I'm trying to understand where it comes from, but all I get is conflicting information.
I think it's easy to find examples of Kodachrome scans with blue casts on the web. Some scans have been corrected, probably scanned with Nikon scanners, but many are uncorrected, yet are disturbingly held out as examples of Kodachrome's great color. When I scan with my Sprintscan 4000, the blue cast is evident with straight, ICC profiled scans.
If you look at the characteristic curves again, the red has the highest density, meaning that less red light is passed through the dyes. The lower density for blue means more blue light is allowed to pass.
I can't comment on the red or yellow casts as I've never seen them. The Silverfast explanation for a yellow caste just seems backward.
-
Ernst: thanks for the various links. I didn't realise the Coolscan 5000 could multiexpose and thus increase the dynamic range. Nice facility to play around with for single scans, but would cause a significant increase in time and probably be not that useful for me given the large number of slides I want to scan. I'm also a bis sus about multiscans because of the chance of additional blurring caused by the extra passes.
Both in Vuescan and Silverfast there are two options for more film scanners: Multi Sampling which uses one scan run but takes more samples per scan step. Multi Exposure scan (or whatever it is called) that is done with two scan runs, one with longer exposure sampling, the other with normal exposure sampling and then the two are combined with a smart algorithm. The first will not produce bad register, the second could but may be acceptable with 35mm scans. The first can reduce noise and by that increase the dynamic range a bit into the shadows, the second is usually a further compromise between noise and extended dynamic range. I think Vuescan can use both simultaneously, don't count the hours.
I wonder whether the discussed blue bias of Kodachrome isn't a sign of Kodachrome slides better fade resistance versus all the other chromogenic process slides. One may wonder whether fresh Kodachrome<>Ektachrome slides are much closer in their color representation if suitable scanner profiling is used for both. What is seen on the web is most likely not a good source to make judgements on. Jeff's comment on the other observer (scanner sensor RGB filters) versus the human eye/brain still stands. For slide projection/human observation Kodak must have set the same goals for what is good color but that doesn't translate to scanning. Another factor could be the Kodachrome's complex layer structure. ICE in its first version couldn't be used for B&W and Kodachrome, the last has changed though in later versions. It seldom happens that I scan slides these days so I have no proof of this speculation.
met vriendelijke groeten, Ernst Dinkla
Try: http://groups.yahoo.com/group/Wide_Inkjet_Printers/
-
I wonder whether the discussed blue bias of Kodachrome isn't a sign of Kodachrome slides better fade resistance versus all the other chromogenic process slides.
Are you saying Kodachrome looks blue because in comparison other slide films are yellowish due to fading? I think that's a stretch!
One may wonder whether fresh Kodachrome<>Ektachrome slides are much closer in their color representation if suitable scanner profiling is used for both.
A scanner profile will preserve the blue cast, not eliminate it. The blue cast is shown in the characteristic curves - would Kodak publish the characteristic curves of faded film?.
What is seen on the web is most likely not a good source to make judgements on.
In my previous post I attached 2 images. The first one is the unadjusted scan with a Kodak Q60 Kodachrome target (scarse/argyll) profile applied. The second one is a simulation of the appearance of the first one when projected, using CIECAM02 to model the changes in viewing conditions. The difference in appearance is due solely to the difference in viewing conditions.
Jeff's comment on the other observer (scanner sensor RGB filters) versus the human eye/brain still stands.
I agree with Jeff and was only adding additional considerations to what had already been said.
-
One thing that has not been mentioned here is that Kodachrome film was designed for projection and printing. Ektachrome, on the other hand, was designed for viewing over a 5,000K light source. In other words, think of the blue cast in Kodachrome as a color compensating filter for a tungsten light source.
Bob Rapp
-
And no, you absolutely need to optimize the tone & color as you scan and not try to fix it after the fact in PS.
Jeff, why?
With consumer desktop scanners, all adjustments in scanning software (other than focus, exposure, ICE, multi-sample and multi-pass) merely edit the pixel data derived from the scanner’s CCD.
For purposes of this discussion, a scanner has two essential parts, a light source and a light sensor; e.g. a cold cathode fluorescent lamp (CCFL) light source and a CCD light sensor. For any scanning software to implement a tone or color adjustment at the scan stage, as I think your comment implies, the scanning software would have to be able to control and adjust (1) the spectral output of the CCFL and/or (2) the way the CCD responds when exposed to light. Scanning software can do neither.
Therefore, since tone and color adjustments just edit the pixel data using typical image editing tools, why do such adjustments absolutely need to be optimized using the scanning software rather than after the fact in PS? In both cases, the adjustments are made after the actual scan.
Some people may prefer to make such “software” adjustments with scanning software. I prefer to make such adjustments using Camera Raw and Photoshop because:
1. With most scanning software, the adjustments are baked into the scan output file, and tweaking these adjustments later requires another scan. In other words, the adjustments are destructive. On the other hand, such adjustments in Camera Raw or PS can be done non-destructively and can easily be tweaked later if necessary.
2. I much prefer the tools and capabilities of Camera Raw and PS to those in VueScan, Silverfast or other scanning software I’ve tried.
-
Jeff, why?
With consumer desktop scanners, all adjustments in scanning software (other than focus, exposure, ICE, multi-sample and multi-pass) merely edit the pixel data derived from the scanner’s CCD.
Well, for the same reason that I suggest people optimize tone and color in Camera Raw before opening in Photoshop :~)
To be honest, I have very little experience with consumer grade scanners. I've got one but I only ever use it for reflective scans, never film.
For that I use an Imacon Flextight scanner which does apply edits to the image in the internal color space of the scanner in high (16 bit) before doing the final output file. I can get better results with dense film this way than trying to fix a scan after the fact in Photoshop. The only thing I like doing in the scanner that isn't final is the black and white point clip. So my scans are a bit flat but corrected for tone curve and color in the scanner software then saved using the scanner's profile as an input profile. Upon opening in Photoshop I convert to ProPhoto RGB 16 bit for further edits...
Seriously, if you are trying to scan Kodachrome on a flatbed scanner, I don't think the results will be very good. It's tough to deal with Kodachrome's contrast range on an Imacon Flextight. The best scans I've ever gotten (from 120MM Kodachrome 64) were done on a drum scanner and they had a heck of a time setting up for 120mm Kodachrome (cause, well, it was different than 35mm and their scanner didn't have a decent default starting point).
-
Well, Jeff, we certainly agree about Camera Raw. I used to do everything in PS, until I upgraded from CS to CS4. Now I do most of my processing in Camera Raw. If anyone hasn’t tried processing their film scans in Camera Raw or Lightroom, do yourself a favor and give it a try. Make sure you don’t open an untagged file in Camera Raw (and maybe Lightroom), however, or you won’t get the results you might expect.
Jeff, you also rightly noted the 16 bit versus 8 bit concern, but since most scanners capable of scanning film output 16 bit tiff files, I didn’t get into that concern.
Okay, you get better results using the Imacon software. I can’t argue with that, but I’m puzzled why that would be the case. If you make hardware exposure adjustments in the scanning software, of course that would explain it. On the other hand, if it’s just because you make the edits in the scanner’s internal color space instead of ProPhoto RGB 16 bit in PS, I’m puzzled. You make the conversion to ProPhoto anyway and I would not expect there to be any visible difference if you make the same adjustments in the scanner’s internal color space versus ProPhoto 16 bit in PS.
-
Okay, you get better results using the Imacon software. I can’t argue with that, but I’m puzzled why that would be the case.
Because I think the Imacon software, Flextight, does have more control over the hardware (scan times) and the raw scanned image than consumer level scanner software (although I do use SilverFast on my Epson Perfection V750-M Pro even though I don't use that scanner for film).
To be truthful I haven't visited this subject for a few years since I went through a round of scanning a ton of film for a retrospective print show. All I specifically remember is that doing side by side tests of optimizing the image in Flextight vs. Photoshop resulted in better IQ if I optimized in Flextight first.
On the V750 Epson scanner (which I don't use for film except for the odd 8 x 10 chrome I need a scan of) I would tend to agree that a flat'ish decent tone and color corrected scan can be fixed just fine in Photoshop (except for the odd anti-Newton rings :-).
-
Are you saying Kodachrome looks blue because in comparison other slide films are yellowish due to fading? I think that's a stretch!
A scanner profile will preserve the blue cast, not eliminate it. The blue cast is shown in the characteristic curves - would Kodak
publish the characteristic curves of faded film?.
In my previous post I attached 2 images. The first one is the unadjusted scan with a Kodak Q60 Kodachrome target (scarse/argyll) profile
applied. The second one is a simulation of the appearance of the first one when projected, using CIECAM02 to model the changes in viewing
conditions. The difference in appearance is due solely to the difference in viewing conditions.
I agree with Jeff and was only adding additional considerations to what had already been said.
I mentioned several causes for possible flaws in Kodachrome scanning, adding to the ones already mentioned and I didn't suggest that your plots show fading.
Kodachrome has a reputation of not fading starting with the version after roughly 1938. Dark fading that occured affected the yellow dye though. Ektachrome in all its incarnations dark faded faster and mainly on the cyan layer. Kodachrome didn't suffer of yellow staining, Ektachrome does. Both faded in projection on magenta but the Kodachrome much faster. Wilhelm recommended 1 hr maximum exposure for Kodachrome and 2.5 hrs for Ektachrome, Fujichrome 5 hours. Gathered from Wilhelm's book on the permanence etc of color. You will find some of that in the Wiki pages on Kodachrome. http://en.wikipedia.org/wiki/Kodachrome With the change to digital photography the actual scanning of films today will be more focused on film that has been archived/projected than on fresh film. Kodachrome's use even more shrinking over a longer period. So it is certainly not a stretch.
I do not understand why a Kodachrome slide film should get a blue bias to compensate a yellowish projector lamp and no other slide film needs the same compensation. The same observation is made here:
http://www.apug.org/forums/forum40/81778-love-hate-kodachrome-why-all-passion-here.html
Could be corrected on his assumptions of Kodachrome fade resistance though.
What you see as different plots is set in perspective there too, other dyes, other densities, other spectral plots. Kodachrome has a saturated gamut and that will show. More saturation in blue possible which is something else than a blue bias, the last should show in the neutrals. A richer film like Kodachrome will have more deviations on the neutral axis in any display condition as it is harder to control CMY layers, development, lamp aging etc when color contrasts are higher. Profiling in digitalisation can help then but on average not per slide.
There are articles on the Silverfast site on the Kodachrome profiling aspect: http://www.silverfast.com/show/kodachrome-targets/en.html. Blue is certainly a problem when KC is scanned using an EC profile. See also the scanning part of the wiki Kodachrome page: http://en.wikipedia.org/wiki/Kodachrome. And yes, Kodachrome gamut should be available then but a blue cast is a flaw in a scan workflow.
With Jeff's remark on different observers you can add an extra difference occuring in this discussion: the Sprintscan's white cool cathode fluorescent lamp versus the LEDs of the Coolscan 5000. I have seen and measured some scanner lights (Eye One, Epson V750 cold cathode. Mirage II cold cathode) and they differ in spectral output. The two types of linear sensors will have (smaller) differences in RGB filters. Jeff's Imacon has a (halogen) tungsten lamp if I recall it correctly (edit: wrong, it looks like 5000K fuorescents are used on Imacons) and most likely a Kodak linear CCD sensor very similar to the one in the Nikon 8000. With all systems tweaked for the chromogenic dyes of Ektachrome-Fujichrome you can expect deviations with Kodachrome dyes.
A RAW export of a 16 bit file to a RAW processor is possible with Vuescan too. There is something to be said for editing in scanner space and 16 bit. Of course the file itself isn't comparable to a Bayer sensor RAW file.
Flatbeds crept into this discussion for other reasons, the Nikon Coolscans, the SprintScan, are of another class.
met vriendelijke groeten, Ernst Dinkla
New: Spectral plots of +220 inkjet papers:
http://www.pigment-print.com/spectralplots/spectrumviz_1.htm
-
I mentioned several causes for possible flaws in Kodachrome scanning, adding to the ones already mentioned and I didn't suggest that your spectral plots show fading.
Well, you said "One may wonder whether fresh Kodachrome<>Ektachrome slides are much closer in their color representation if suitable scanner profiling is used for both." From this I understand you to say that the blue cast could be the result of using film that was not fresh, ie faded. If that were the case, then since the Kodak curves also show the blue cast, they would have to be based on non-fresh or faded film, too, otherwise the Kodak curves would not show the cast.
Kodachrome has a reputation of not fading starting with the version after roughly 1938. Dark fading that occured affected the yellow dye though. Ektachrome in all its incarnations dark faded faster and mainly on the cyan layer. Kodachrome didn't suffer of yellow staining, Ektachrome does. Both faded in projection on magenta but the Kodachrome much faster. Wilhelm recommended 1 hr maximum exposure for Kodachrome and 2.5 hrs for Ektachrome, Fujichrome 5 hours. Gathered from Wilhelm's book on the permanence etc of color. You will find some of that in the Wiki pages on Kodachrome. http://en.wikipedia.org/wiki/Kodachrome With the change to digital photography the actual scanning of films today will be more focused on film that has been archived/projected than on fresh film. Kodachrome's use even more shrinking over a longer period. So it is certainly not a stretch.
Thee blue cast of Kodachrome is inherent to fresh Kodachrome. It has nothing to do with the fading properties of other emulsions. I don't have my copy of Wilhelm handy, but if what you say is correct about magenta fading with light exposure, that would create a green cast, would it not? But let's limit the discussion to properly developed and stored Kodachrome scanned using ICC profiles.
I do not understand why a Kodachrome slide film should get a blue bias to compensate a yellowish projector lamp and no other slide film needs the same compensation. The same observation is made here: http://www.apug.org/forums/forum40/81778-love-hate-kodachrome-why-all-passion-here.html
I don't know why either. Perhaps because it makes printing and scanning more difficult?
With Jeff's remark on different observers you can add an extra difference occuring in this discussion: the Sprintscan's white cool cathode fluorescent lamp versus the LEDs of the Coolscan 5000. I have seen and measured some scanner lights (Eye One, Epson V750 cold cathode. Mirage II cold cathode) and they differ in spectral output. The two types of linear sensors will have (smaller) differences in RGB filters. Jeff's Imacon has a (halogen) tungsten lamp if I recall it correctly and most likely a Kodak linear CCD sensor very similar to the one in the Nikon 8000. With all systems tweaked for the chromogenic dyes of Ektachrome-Fujichrome you can expect deviations with Kodachrome dyes.
But all of these deviations are compensated with a proper ICC profile.
Here is an experiment to test the theory that the blue cast is due to the curves. The first graph shows the Kodachrome 25 curves digitized and graphed in Excel.
The second graph shows the same K25 curves after the Red density is scaled by 0.84 and the Green density is scaled by 0.96. These scaling serve to align the three cuves.
Next is a scan with Kodachrome ICC profile applied and converted to sRGB without any adjustments. The last image has been adjusted (in the profile space) in levels with Red midtone level (gamma) set to 1.19 (1/0.84) and the Green midtone level set to 1.04 (1/0.96) - in other words, the same adjustments that aligned the curves in the Excel model. The last image has also been reduced in contrast with an RGB midtone level adjustment of 1.25. No other adjustments have been made. The adjusted image is warmer and the excess blue in the whitewashed barn wall is removed along with excess blue in the shingled roof of the shed on the right. Certainly the curve-aligned image is without the blue cast and is a better starting point for additional editing - and a better representation of the image as it would appear projected.
-
There is a good explanation at that link why both Kodachrome and Ektachrome come close to one another in projection despite different curves for RGB dyes. Call it a metameric match in projection light. The curves you see are curves for different dyes, it isn't a different dose of the same dyes creating the curve displacement.
A quote from that page:
If you compare the first figure to the third figure, you can compare the Ektachrome with Kodachrome. You can immediately see that the curves of the 3 Kodachrome dyes are NOT matched, and that the Cyan is considerably higher than the other two. As a result, a visual neutral may look cyanish. To see why I say “may” look at the 4th figure which is a unit neutral of the Kodachrome dye set. You can immediately see that the cyan dye requires more density to give a visual neutral of 1.0 due to the fact that the cyan dye is very narrow in absorption. It is also low in unwanted densities and this introduces a big dip or gap in the neutral at about 600 nm. The lower unwanted absorption of the cyan in the green and blue region of the spectrum are responsible in part, for the improved overall color saturation.
The result of all of this though is that Kodachrome K-14 products have slightly more illuminant sensitivity than the E6 product family, but since they are all intended to be projected, who cares. It also means that even under identical illumination, some people may view Kodachrome slides differently than others, due to the slight, but normal differences in the color receptor pigment balance in individual eyes. On the other hand, Kodachrome colors have a vibrant saturation with less impurities for this same reason, but in some cases, that advantage is scene dependant.<<
end of quote
Call the blue shift in scanning a metameric failure due to changed scanner lighting and another observer (scanner sensors) and I can agree. The blue shift isn't Kodachrome color and isn't there for projection light compensattion.
Yellow fading in dark storage will result in a blue cast. Cyan fading of Ektachrome will result in an orange/red cast combined with yellow staining more orange. In short the two drift apart, one to the blue the other to orange. Any projection will cause magenta loss and an additionally shift, Kodachrome to bluegreen and Ektachrome to a dirty orange.
With the Kodachrome shift to blue caused by the scanner hardware (though that can vary per scanner as I already mentioned) any fading of both Kodachrome (blue-bluegreen) Ektachrome (red/orange-dirty orange) will at least increase the shift to blue already created by the scanner hardware. A fresh Kodachrome and Ektachrome IT8 calibration target may help with fresh slides to get a better match, faded slides will not match as nice and set both film types more apart. The other variable is of course fading of the targets but we are professionals so we have fresh targets :-)
met vriendelijke groeten, Ernst Dinkla
Try: http://groups.yahoo.com/group/Wide_Inkjet_Printers/
-
Call the blue shift in scanning a metameric failure due to changed scanner lighting and another observer (scanner sensors) and I can agree. The blue shift isn't Kodachrome color and isn't there for projection light compensattion.
Ernst, you're a hard person to convince.:)
The blue shift in scanning is due to the fact that most scanners don't use tungsten illuminants at 3200K, and when using profiles, full adaptation is made to D50.
With the Kodachrome shift to blue caused by the scanner hardware (though that can vary per scanner as I already mentioned) any fading of both Kodachrome (blue-bluegreen) Ektachrome (red/orange-dirty orange) will at least increase the shift to blue already created by the scanner hardware. A fresh Kodachrome and Ektachrome IT8 calibration target may help with fresh slides to get a better match, faded slides will not match as nice and set both film types more apart. The other variable is of course fading of the targets but we are professionals so we have fresh targets :-)
I say again, any blue shifts caused by the hardware will be corrected by the ICC profile. Yet, a blue cast remains even with a profile, because the profile ensures that the inherent blue cast is accurately reproduced.
One more quote from Hunt's definitive reference, The Reproduction of Colour, pgs. 229-230. If this doesn't convince you, I don't know what else to say:
In figure 14.9 are shown typical sets of characteristic curves of red, green, and blue integral density plotted against log exposure for a reversal (a) and two print(b) and (c) materials, when the colour of the exposing light is such that most exposure levels are reproduced so as to appear grey in the viewing conditions typical of those used in practice for each material. When the viewing conditions consist of projection by tungsten light in a darkened room, the light from the projector appears yellowish (Hunt, 1965), and therefore to obtain results that appear grey the picture has to be slightly bluish (see Section 5.7 [quoted previously]); this is why the curves of Fig. 14.9(a), which relate to materials for tungsten-light projection, are not even approximately coincident, the blue densities being lower than, and the red densities higher than, the green densities, in order to produce the bluish result required...
Figure 14.9(a) is attached below. The caption reads:
Fig. 14.9. Typical characteristic curves of integral density plotted against log exposure when the colour of the exposing light is such that most exposure levels are reproduced grey. (a) Reversal colour film (such as Kodachrome film) for making camera pictures for projection by tungsten light...
Not only does Hunt state that it is the case for Kodachrome, the example curves are essentially the same as the K25 curves posted previously.
-
For anyone coming to this point who wants a summary, I have started a new thread here:
http://www.luminous-landscape.com/forum/index.php?topic=50387.0
-
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.
-
Guy, your explanation has convinced me that my suspicion that you may be a shill was unfounded. I apologize.
-
When printing with the Dye Transfer Process, we used Principle Masks to lower the contrast range of the slide from around 2.9 down to around 2.1, but in the process we also removed large, known distortions in the original slide dyes. For example, an Ektachrome cyan dye absorbs 100% of red light, just as it should do, but it also absorbs 30% of green light. This means that the cyan dye also contains 30% magenta that does not belong there. If we shot a principle mask through a red filter and develop it to 30% of the contrast or gamma that is needed for a separation, then put that on top of the raw film when the separation shot through the G61 green filter is made (the one used to expose the magenta matrix), then everywhere cyan and magenta both appear in the image 30% of the exposure will be blocked in the magenta matrix. This way the 30% false magenta in the original cyan dye in the slide can be eliminated. A cyan absorption in the Magenta dye was treated the same way.
I regularly made two sets of red separations, one with an R29 filter and one with an R25 (most people used one R25 separation). By changing the percentages of which separation was given more time on the split exposures, I could make the balance in the greens in the print warmer or cooler, as this varied the curve of the cyan. There was probably also a difference in the exposure of the principle masks used when printing Kodachrome vs Ektachrome and other E6 films, but I do not remember exactly. I could dig for it in the attic if anyone is interested.
Here is a simple account followed by an explanation in more detail about how Dye Transfer was done, and why it produced superior colour rendition over all other print processes at the time:
http://avalon.unomaha.edu/afghan/techinfo/TECHINDX.HTM
-
There was probably also a difference in the exposure of the principle masks used when printing Kodachrome vs Ektachrome and other E6 films, but I do not remember exactly. I could dig for it in the attic if anyone is interested.
Yes, that would be interesting. I'm wondering if there is also a difference in development to alter the contrast of the masks?