Ad
Ad
Ad
Pages: « 1 2 3 [4] 5 »   Bottom of Page
Print
Author Topic: Kodachrome Revisited  (Read 15926 times)
davidh202
Sr. Member
****
Offline Offline

Posts: 562


« Reply #60 on: January 27, 2011, 09:06:58 PM »
ReplyReply

Thanks Mark,
I have found and bookmarked your site and articles, Will Read For sure!.
I always wondered why my Microtek i900 came with a It8 Ektachrome target but not a Kodachrome.I imagine now, that because of the complexity of getting good results they just decided to completely avoid the issue
There was no mention in any of the literature of having to use a Kodachrome target for Kodachrome. I had assumed when I got the scanner that a slide was a slide. It  wasn't untill I got deeper into color management and profiling recently that I have been enlightened.
Logged
crames
Full Member
***
Offline Offline

Posts: 210


WWW
« Reply #61 on: January 27, 2011, 10:00:59 PM »
ReplyReply

I still have to think the real reason for the bluish cast to many Kodachrome scans is simply that most scanners see and interpret the K-14 dyes differently than they see E-6, and that most scanners have input profiles based on E-6 pigments only makes it all the more apparent.

That's definitely the favored theory around here. To test it, here is a scan of a Color Checker and gray scale on K64, from a SprintScan 4000.

The first frame has an Ektachrome profile.

The second frame has a Kodachrome profile.

The third frame is the second one with Kodachrome profile again but with the addition of a simple Levels mid-tone adjustment of Red 1.09 and Blue 0.8 to neutralize the gray-scale somewhat and provide a reference.

The surprising thing is that the excess blue is present to the same extent with either profile. There is no increase of blue with the Ektachrome profile. The difference is a weaker red response with the Ektachrome profile, resulting in an additional cyanish cast.

So you could say that this scanner has a cyanish cast due to the way it sees and interprets the Kodachrome dyes, but that is corrected by a Kodachrome profile. There remains a separate cast from excess blue that is not corrected by any profile because it is a part of the film image itself, hence the need for special Kodachrome modes, etc., etc.
Logged

Cliff
tgray
Newbie
*
Offline Offline

Posts: 39


« Reply #62 on: January 28, 2011, 09:41:36 AM »
ReplyReply

So you could say that this scanner has a cyanish cast due to the way it sees and interprets the Kodachrome dyes, but that is corrected by a Kodachrome profile. There remains a separate cast from excess blue that is not corrected by any profile because it is a part of the film image itself, hence the need for special Kodachrome modes, etc., etc.

I always interpreted that the special Kodachrome modes try to correct for the same thing that is mostly corrected by a profile.  A lot of people use scanners and never profile them.  So a Kodachrome mode takes care of most of the misreading of the cyan dye, which would presumably be caught with a good profile as well.
Logged
Tim Lookingbill
Sr. Member
****
Offline Offline

Posts: 1230



WWW
« Reply #63 on: January 28, 2011, 11:06:24 AM »
ReplyReply

All I want to know is why the colors in crame's scans of Kodachrome look so different from pfigen's with regards to density, contrast and richness in color especially in the skin tone patches and yellow, red and blue.
« Last Edit: January 28, 2011, 11:10:10 AM by tlooknbill » Logged
crames
Full Member
***
Offline Offline

Posts: 210


WWW
« Reply #64 on: January 28, 2011, 11:40:23 AM »
ReplyReply

All I want to know is why the colors in crame's scans of Kodachrome look so different from pfigen's with regards to density, contrast and richness in color especially in the skin tone patches and yellow, red and blue.

Because I am showing in the first two frames exactly what comes out of the scanner without any manipulation. The only difference is the ICC profile applied. This was to show what an ICC profile does correct and also what it doesn't correct. (The third frame is nothing more than a gray balance based on using gamma adjustments to align the RGB curves - notice that with this simple adjustment the color of the skin patch, at least, is right on the money.)

In contrast, pfigen has described that his scans are not exactly what comes out of the scanner - they have been skillfully enhanced to bring out the Kodachrome look.
Logged

Cliff
crames
Full Member
***
Offline Offline

Posts: 210


WWW
« Reply #65 on: January 28, 2011, 12:34:24 PM »
ReplyReply

I always interpreted that the special Kodachrome modes try to correct for the same thing that is mostly corrected by a profile.  A lot of people use scanners and never profile them.  So a Kodachrome mode takes care of most of the misreading of the cyan dye, which would presumably be caught with a good profile as well.

I can imagine a special Kodachrome mode that includes a profile, but otherwise, the special mode needs to do something different from what a profile does.

After all that has been presented in these three Kodachrome threads, do you think that the famous blue cast is simply due to misreading the cyan dye?
Logged

Cliff
Tim Lookingbill
Sr. Member
****
Offline Offline

Posts: 1230



WWW
« Reply #66 on: January 28, 2011, 01:02:16 PM »
ReplyReply

Quote
In contrast, pfigen has described that his scans are not exactly what comes out of the scanner - they have been skillfully enhanced to bring out the Kodachrome look.

I believe you may have misread what pfigen said. He said the scans he just posted match exactly what he sees on his light table. He's not bringing out anything that isn't in the slide.

Do your scans you just posted match what you see on your light table or however you view your Kodachrome slides? And if so, which of the three match?
Logged
tgray
Newbie
*
Offline Offline

Posts: 39


« Reply #67 on: January 28, 2011, 01:08:36 PM »
ReplyReply

After all that has been presented in these three Kodachrome threads, do you think that the famous blue cast is simply due to misreading the cyan dye?

I get the feeling that it plays a large role in it.  To the eye, Kodachrome and E6 both display pretty neutral, regardless of light source.  So I do think it's mostly the interaction of the RGB filters in the scanner and the spectral curves of the dyes that make Kodachrome scan differently than E6.  The 3200/5000 balance might play a part or might not, I don't know.  Even if it does, I think the cyan dye is a major aspect.  It makes physical sense to me and is also supported by ex-Kodak guys.

Frankly I'm not very impressed with the Kodachrome mode of my Coolscan when using Nikonscan.  I got better results ignoring the mode and adding a preset layers adjustment to shift the colors around a bit. And I recently got MUCH superior colors by using a profile I generated using a target.  On the other hand, most E6 scans quite easily and much closer to the original without using a profile.
« Last Edit: January 28, 2011, 01:10:26 PM by tgray » Logged
pfigen
Sr. Member
****
Offline Offline

Posts: 469


WWW
« Reply #68 on: January 28, 2011, 01:26:07 PM »
ReplyReply

I took a quick little shot at your MacBeth chart, adding a separate Curves Adjustment Layer to both the first and the second images, concentrating more on the MacBeth chart than your gray card, which seems to be a bit cooler than the MacBeth. Both versions that I did, which are fairly close to each other, put the MacBeth as viewed on my Sony Artisan, much closer to the actual MacBeth I have sitting right next to me. The Curves were much more complicated than "normal" curves, having usually seven or eight adjustment points in each channel to really fine tune the respective gray patches. While the Kodachrome input profile does indeed improve the out of scanner rendition on this particular scanner, it's also not that difficult to tweak the original file to more or less match. It also illustrates that the more or less global adjustments done on the far right image were maybe less effective than they could have been. Kodachrome is always an interesting exercise, but one that never fails to satisfy.
Logged
Tim Lookingbill
Sr. Member
****
Offline Offline

Posts: 1230



WWW
« Reply #69 on: January 28, 2011, 02:28:50 PM »
ReplyReply

Peter, so you're making crame's scan match to an actual CC chart next to you?

I don't see the point in doing this.

I would think you'ld want to construct edits to make crame's scan match to your own Kodachrome slide of the CC chart.

I wasn't under the impression from the look of your scans that Kodachrome reproduces reality all that accurately. It adds its own color patina to the recording of reality that is unique and all its own.

Maybe I'm missing something here.
Logged
crames
Full Member
***
Offline Offline

Posts: 210


WWW
« Reply #70 on: January 28, 2011, 03:22:59 PM »
ReplyReply

I believe you may have misread what pfigen said. He said the scans he just posted match exactly what he sees on his light table. He's not bringing out anything that isn't in the slide.

Yes, he said "The image I posted was VERY close to how the slide looked on a Just-Normlicht D-50 lightbox." But what comes out of the scanner isn't what you see on a light table (or projector). The scan has to be edited to make it look that way. dfigen is around so hopefully he will correct me if I am misstating.

Quote
Do your scans you just posted match what you see on your light table or however you view your Kodachrome slides? And if so, which of the three match?

In terms of hue and chroma, the closest is definitely the third, but the actual slide looks a lot brighter and the grays are more neutral looking, but I'm comparing a D50 light box to a D65 calibrated monitor, so that muddies the water about neutrality.

The purpose of that image was only to show the visible effects of two profiles. I made no effort to adjust the brightness or anything else to match the lightbox as that would have obscured what the profiles are doing. I only made a rough gray balance in the third panel and no other.

The second one has an accurate Kodachrome profile that corrects for the light source and filters of the scanner, yet doesn't come close to matching the appearance of the slide.
« Last Edit: January 28, 2011, 05:21:58 PM by crames » Logged

Cliff
crames
Full Member
***
Offline Offline

Posts: 210


WWW
« Reply #71 on: January 28, 2011, 04:17:59 PM »
ReplyReply

I get the feeling that it plays a large role in it.  To the eye, Kodachrome and E6 both display pretty neutral, regardless of light source.  So I do think it's mostly the interaction of the RGB filters in the scanner and the spectral curves of the dyes that make Kodachrome scan differently than E6.  The 3200/5000 balance might play a part or might not, I don't know.

If you don't use a profile, you can get all kinds of color problems. But once you use an accurate ICC profile, you are getting the color as seen by the CIE Standard Observer, with a perfect D50 light source. A profile invalidates the argument that the blue cast is due to the scanner filters or light source. (Aren't there any color management gurus around to weigh in on this, or have they all left the building?)

Quote
Even if it does, I think the cyan dye is a major aspect.  It makes physical sense to me and is also supported by ex-Kodak guys.

I've quoted an ex-Kodak guy a few times (RWG Hunt, The Reproduction of Colour) about the nature of the blue cast in Kodachrome:
 
Quote
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...

Regarding the cyan dye, I could be wrong but looks to me as though the excess blue is mainly controlled by the yellow dye, if you were to look at the characteristic curves.

Quote
Frankly I'm not very impressed with the Kodachrome mode of my Coolscan when using Nikonscan.  I got better results ignoring the mode and adding a preset layers adjustment to shift the colors around a bit. And I recently got MUCH superior colors by using a profile I generated using a target.  On the other hand, most E6 scans quite easily and much closer to the original without using a profile.

Have you tried the new profile and the Kodachrome mode together?
Logged

Cliff
crames
Full Member
***
Offline Offline

Posts: 210


WWW
« Reply #72 on: January 28, 2011, 05:20:11 PM »
ReplyReply

I took a quick little shot at your MacBeth chart, adding a separate Curves Adjustment Layer to both the first and the second images, concentrating more on the MacBeth chart than your gray card, which seems to be a bit cooler than the MacBeth. Both versions that I did, which are fairly close to each other, put the MacBeth as viewed on my Sony Artisan, much closer to the actual MacBeth I have sitting right next to me. The Curves were much more complicated than "normal" curves, having usually seven or eight adjustment points in each channel to really fine tune the respective gray patches. While the Kodachrome input profile does indeed improve the out of scanner rendition on this particular scanner, it's also not that difficult to tweak the original file to more or less match. It also illustrates that the more or less global adjustments done on the far right image were maybe less effective than they could have been. Kodachrome is always an interesting exercise, but one that never fails to satisfy.

I made no attempt to match the real Color Checker, but I do think that the two small gamma adjustments make a large improvement. It's something quick and simple, doesn't require a reference, and looks a lot better than the uncorrected blue stuff. The gamma adjustment only tilts the red and blue curves, without changing their overall shapes. The idea is to get rid of the blue, simply while minimally affecting other aspects of the image.

By editing to match the real Color Checker aren't you in effect turning the KC tone curve into a straight line and removing part of the look? Sorry, it's been a long, tiring Friday and I'm not getting it.

Edit: Oops - I see that you didn't straighten the curves.
« Last Edit: January 28, 2011, 11:03:42 PM by crames » Logged

Cliff
tgray
Newbie
*
Offline Offline

Posts: 39


« Reply #73 on: January 28, 2011, 05:48:45 PM »
ReplyReply

If you don't use a profile, you can get all kinds of color problems. But once you use an accurate ICC profile, you are getting the color as seen by the CIE Standard Observer, with a perfect D50 light source. A profile invalidates the argument that the blue cast is due to the scanner filters or light source. (Aren't there any color management gurus around to weigh in on this, or have they all left the building?)

Why exactly?  I'm not getting a blue cast when I use a profile...

Quote
Regarding the cyan dye, I could be wrong but looks to me as though the excess blue is mainly controlled by the yellow dye, if you were to look at the characteristic curves.

Sorry.  I had a brain fart there.  I still think this statement is accurate: "So I do think it's mostly the interaction of the RGB filters in the scanner and the spectral curves of the dyes that make Kodachrome scan differently than E6."

Throw in a different color balance on top of that as you quoted from hunt, and you'll get all kinds of funky casts.
Logged
crames
Full Member
***
Offline Offline

Posts: 210


WWW
« Reply #74 on: January 28, 2011, 07:21:47 PM »
ReplyReply

Why exactly?  I'm not getting a blue cast when I use a profile...

Please clarify: you don't get a blue cast when you use a profile by itself, or you don't get a blue cast when you use a profile along with the Kodachrome setting?
 
Quote
I still think this statement is accurate: "So I do think it's mostly the interaction of the RGB filters in the scanner and the spectral curves of the dyes that make Kodachrome scan differently than E6."

Throw in a different color balance on top of that as you quoted from hunt, and you'll get all kinds of funky casts.

That's not what Hunt says. He says the Kodachrome is manufactured to have a blue tint. The blue tint that has been manufactured into the film has nothing to do with the scanner, the light source, the filters. It's not a different color balance. It's not removed by a filter or by changing the light source. You can see it in the characteristic curves that Kodak provides.

A gray scale shot on Kodachrome has increasingly more blue as you go from white to black. There is also decreasing red as you go from white to black. The scanner will pick this up.

Now if you scan without an ICC profile, the characteristics of the particular scanner can introduce other things that add to the blue tint that Hunt is talking about. I just showed an example where my scanner can add a cyanish cast if I'm not using the right profile. With the Kodachrome profile in place, there is no cyanish cast, only the real blue tint that the scanner has measured off the film.

Think about what an ICC profile is supposed to do. It translates the device-dependent RGB values of a scanner into a device-independent representation. The lookup table in a scanner profile compensates for the light source, the filters, the dyes, the sensor, etc. So if there is a defect in the color being caused by those things, the profile will remove it. The profile won't/shouldn't remove any of the actual color of the film, such as the blue tint described by Hunt.
Logged

Cliff
Gordon Buck
Sr. Member
****
Offline Offline

Posts: 409



WWW
« Reply #75 on: January 28, 2011, 07:41:10 PM »
ReplyReply

I've been reading these Kodachrome threads with much more interest than understanding; however, my lack of expertise does not prevent me from contributing to the message thread because I too have shot the "last" roll of Kodachrome -- and the Tee shirt from Dwaynes is on its way to me!

My last roll of Kodachrome was shot, appropriately, in my old Konica T2.  At the same time, I (sometimes) grabbed similar shots with a Canon G9 digicam.  I had Dwaynes scan the roll on the assumption that they knew how to do it.  To my surprise, the Dwaynes scans came back on a Kodak Picture CD and I did not like them.  My next step was to scan the slides myself using a Canon FS4000 scanner using Vuescan.  I set up Vuescan for Kodachrome slides with no additional adjustments (no white balance, levels, etc.).  The other part of my plan was to process the G9 shot as a "Kodachrome" by using the Pixel Genius PhotoKit Color2 setting as"Classic Chrome".

The scans were capture sharpened with Pixel Genius Capture Sharpener 2, reduced in size and then output sharpened with Output Sharpener 2.  The G9 was converted in ACR in my relatively simple and consistent manner, tweaked as Classic Chrome, reduced in size and output sharpened.

To my eye -- and surprise -- the Dwaynes scan most resembles the real slide but is relatively low resolution, 8 bit and seems to be highly compressed.  If I were to make a print from the slide, I'd use the Vuescan version with additional post processing.  On the other hand, if I wanted a print of the scene, I'd take another hack at the G9 RAW file.

Anyway, all this means something or other, I'm sure.


Logged

davidh202
Sr. Member
****
Offline Offline

Posts: 562


« Reply #76 on: January 28, 2011, 09:26:56 PM »
ReplyReply

 I know that we are currently speaking about scanning, but now that I think about it...
In the early days when sorting a new batch of slides, I used a very inexpensive folding light tray that used a standard tungsten light bulb, I knew nothing of D50 standards back then.Not only that but I have no idea now what the kelvin temp (white balance),  was of the ELH halogen bulb in my Carousel projector,  and I'd venture to say that the different bulbs used by different projectors from various manufacturers varied in color temp also.
    To add to the variables there were projection screens that had a silvered gray surface and some that had a white beaded surface.
I'd imagine that any different combination of these values affected the appearance of the slides when being viewed !?
   Could we not say that each individual situation would render a different appearance when viewed, and some were better than others as far as being true to the Kodachome "look", or true to what we actually remember the scene to have been.
I had said in an earlier post that there are so many variables involved, seems like nothing has really changed much.
   I am curious as to how did one (profile), deal with filtering for the blue cast of Kodachrome, when using a direct print method such as Ilfochrome  (Cibachrome), before the computer age?
 
« Last Edit: January 28, 2011, 09:36:31 PM by davidh202 » Logged
Tim Lookingbill
Sr. Member
****
Offline Offline

Posts: 1230



WWW
« Reply #77 on: January 28, 2011, 10:05:53 PM »
ReplyReply

We also haven't ruled out processing induced differences that may be influencing the different looks of those posting their Kodachrome scans so far.

It would really help and reduce confusion to just post what the slide exactly looks like which should be of scans of scenes that contain a wide range of real world colored objects containing a wide range of different colors lit under real world lighting, otherwise there's no point in profiling if you're not going to show exactly what the scan looks like.

pfigen's scans look so different from everyone else's so far and it's further confusing the issue of establishing the look of Kodachrome. And maybe that's what this discussion will finally prove and rule out which may be processing and/or type (K25, K64, etc.) that's causing the differences. It doesn't matter if the scanner sees blue, cyan. Is it on the actual slide? And if so, why isn't it on other contributor's slide?

Looking at an it8 target or any color target with or without a profile isn't going to tell us how Kodachrome renders every possible real world lighting situation and how Kodachrome reacts to it with every possible color under the rainbow. That's why I asked pfigen if that yellow on that building was truly golden yellow or lemon yellow. Was that white really a pastel pinkish orange under that golden sunset?
Logged
pfigen
Sr. Member
****
Offline Offline

Posts: 469


WWW
« Reply #78 on: January 29, 2011, 12:21:39 AM »
ReplyReply

Tim,

The yellow was definitely not lemon yellow but much more on the golden side, and the white at the top was more pinkish, but all those things can be affected by the emulsion lot, the processing, the time of day, whether or not you had any sort of filter on the lens, whether you shot with a Leica, Nikon or Canon lens and probably about thirty more factors I can't think of at the moment.

The reason I did the corrections on the MacBeth chart previously posted were just to illustrate that there are even more variables to consider when considering who is doing the scanning. The first frame was obviously way too cool, and while the second frame, with the profile applied was considerably better, it too was not only cool but non-linear in the gray balance from light to dark. The third was sort of a half hearted attempt to warm it up a bit, but had a lot of shortcomings as well. This still begs the question of how close to the original color checker was that piece of film? If it pretty much matched, then the K-14 profile wasn't all that good to begin with and with a couple of minutes with well places curves, you could actually better the profile manually. It was all to illustrate that there's more than one way to get this done effectively. If you've got a piece of film with a color checker on it and know how that color checker is supposed to look, do you correct the scan to do that or do you make your scan match the film as close as possible?

If in fact you are using the color checker for some sort of control, then you would probably want to correct it so it looks as close to the real life checker as possible, overriding whatever the profile is giving you straight out of the scanner. The profiled scan is really just a starting point and any good scanner operator will always set white and black points at a bare minimum, but also consider gray balance and selective color as well, where applicable. You just have to examine the pixel values in the scan and see what makes sense. I didn't know off the top of my head what the actual pixel values were for an idealized color checker, but I shot for something like the black patch at around 16-17 and the white patch at around 237-8 or so, while making the rest as neutral as possible. That made all the other colors come in a lot closer all on their own. It's all just the basic digital color correction theory 101 that I've been practicing for the last fifteen years.
Logged
Tim Lookingbill
Sr. Member
****
Offline Offline

Posts: 1230



WWW
« Reply #79 on: January 29, 2011, 12:45:01 AM »
ReplyReply

Quote
The yellow was definitely not lemon yellow but much more on the golden side, and the white at the top was more pinkish, but all those things can be affected by the emulsion lot, the processing, the time of day, whether or not you had any sort of filter on the lens, whether you shot with a Leica, Nikon or Canon lens and probably about thirty more factors I can't think of at the moment.

Then Kodachrome could look like anything no matter what subject and lighting you shoot.

So there's no nailing down the look of Kodachrome. I don't know what else to conclude.

Folks are just seeing what they want to see in this film. Nice bright colors according to Paul Simon is good enough description I guess.
Logged
Pages: « 1 2 3 [4] 5 »   Top of Page
Print
Jump to:  

Ad
Ad
Ad