Color Bias in characteristic curve in negative films

[Photo Engineer]

The negative does all of the heavy lifting here, and the paper just reproduces what the negative has produced. There is no color correction in paper. It is all done in the film.

PE

 

[Mr Bill]

The question is really why the gamma bias exists in the film, and not in the paper, and the correct answer is not because of the mask,...

I presume that when you say "the gamma bias" in the film you are referring to the published film characteristic curves not being exactly parallel to each other.

I was perhaps not emphatic enough earlier (post #8) saying, "If you are using characteristic curves, these are mostly made with a certain densitometer response (status M, I think, for color neg). These use a narrow spectral response, and almost certainly are not how a paper "sees" them - the results are usually broadly similar, but nothing like exact."

This is most likely the major source of your issue.

If you prefer to hear it from a more authoritative source, try Giorgianni and Madden's " Digital Color Management: Encoding Solutions." In the original 1998 version, chapt 7 shows an example set of curves comparing status M densitometer results vs "printing density." (Printing density is how the film "looks" to the paper. ) They're significantly different, with the "printing density" curves all parallel, but the Status M curves clearly not so.

I don't know any simple way for you to get a "printing density." The paper essentially "looks" at a broader spectral section of the film's transmission than what Status M sees, and this is "weighted" by the spectral sensitivity of the paper at each wavelength. Since you already have your own software in place, and presumably are using "curves" that you measured from the published characteristic curves, you might try just using the overall average for all three curves.

 

[Ted Baker]

presume that when you say "the gamma bias" in the film you are referring to the published film characteristic curves not being exactly parallel to each other.

Yes

I was perhaps not emphatic enough earlier (post #8) saying...

I agree that's a possibility, but there are others, one idea that I "like" is the bias is because the "colorspace" of the film is not what you might call well behaved, because of the chromatic co-ordinates of the primaries(dyes) if I use some digital terminology (well behaved meaning gray/white is xR = yG = zB, with x,y,z being some constants). In contrast the "colorspace" of the paper appears well behaved, but I also have problems with that idea...

Teach me to ask difficult questions

If you prefer to hear it from a more authoritative source

Unfortunately I don't think I am going to get a simple answer, however none of this thread goes to waste! Indeed I am using trial and error with splines/"curves" to invert the negative, however a more clear theoretical understanding can stimulate different trials with less errors, and more success.

I am making some progress, watch this space

 

[Photo Engineer]

The book by Ed. Giorgianni et. al. should be quite good. Ed was one of EKs top experts in this.

PE

 

[alanrockwood]

The book by Ed. Giorgianni et. al. should be quite good. Ed was one of EKs top experts in this.

PE

PE,

Where does the book "The Reproduction of Colour", fifth edition by Dr. D.W.G. Hunt rank in the pantheon of books on the theory of color photography. Unlike the other books, I have a copy of that book.

 

[alanrockwood]

You know, some information that would be extremely useful to have (in addition to RGB sensitometry curves) would be spectral response curves for the layers in color films and also spectral absorbance curves for the three layers. I have searched high and low for this information and have only be partially successful in finding it. I assume it mostly falls under trade secret.

It would be possible to measure this information if someone has the right sort of equipment. What would be required would be a tunable source of narrow band light (e.g. a slit spectrometer) to expose the film and a spectrometer to read the developed film. another possibility for exposure might be a set of narrow band light emitting diodes, but that would not be quite as good. Also, the spectrometer responsible for exposing the film (I will call it "excitation") would need to have the capability to balance the power across all wavelengths.

The idea would be to map out the response of the film at each excitation wavelength. The result would be a two dimensional matrix of spectrometric data, i.e. the density of the film for every combination of excitation wavelength and and absorbance at each wavelength of the developed film.

One could then do a principle component analysis (possibly using singular value decomposition or other means) to effectively deconvolute the information from the three layers.

 

[Photo Engineer]

I have not read Ed's book. I was fortunate to interact with him on a personal basis before the book was written, but I have not read the book. As I said in my post "it should be....". Bob Hunt's book was used as a text book and he taught color reproduction at Kodak in our internal "school".

I would rank Evans et. al. at the top for starters and then the other books either tied or ranked, and best ranked by someone who has read both.

The spectrophotometric curves of the dyes are published, as are the spectral responses of the layers. Some of that is posted above and some in the Kodak pdfs.

PE

 

[Mr Bill]

The book by Ed. Giorgianni et. al. should be quite good. Ed was one of EKs top experts in this.

PE

Yes, imo it's a great introductory book, "introductory" meaning different things to different people. At the place where I worked, a large user of photo materials, we were doing a large in-house software project with digital imaging. The IT people found the book long, tedious, and nearly incomprehensible. When I borrowed their book it was a couple of long (but enjoyable) evenings of reading, albeit largely superficial as I skipped over most of the math and the appendices. Afterwards I had to get my own personal copy.

It uses a number of pages on the Kodak Photo CD and its color space and other things that are sort of outdated, which would be my biggest complaint today. However it was updated ~2009. I don't know if he had any direct input, not having seen the book, but Michael Kriss was the series editor for Wiley for this; I nearly got the book for this reason, but was working in other areas by then.

I would rank Evans et. al. at the top for starters and then the other books either tied or ranked, and best ranked by someone who has read both.

I've never seen Evans, but between Hunt and Giorgianni, I'd just say they are very much different. Hunt covers the basic principles more thoroughly, with very little about digital in my copy (5th Ed, 1995 I think; I didn't check). Giorgianni briefly covers the fundamentals then gets into principles of a reproduction system, with plenty of spectral graphs as well as some log luminance vs density of systems. I'm pretty comfortable with such graphs, so they get the ideas across quickly. But if such graphs are foreign to one, it may be a difficult slog; I dunno.

I don't think either one will have much direct application today, at least my editions. But they can greatly further one's understanding. If I could keep only one, it would be Hunt - it covers a broader range. As a note I once met Dr. Hunt at a color conference (early 2000, I'd guess); I told him his book had really started me into the "study" of color. He asked me, "Which book?" I had forgotten that he has another, more specialized book, "Measuring Colour."

I would recommend for most people to get the book(s) from a library first, to see if they suit you. Hardly anyone else I worked with had any interest in either book. I think you have to crave a deeper understanding, as opposed to practical use, for these books. Just some opinions.

 

[RPC]

I would agree about the "introductory" or starter viewpoint comment. I bought the Evans/Hanson/Brewer book a few years back but it turned out, for the most part, to be a little too involved for a hobbyist like myself, although some of it was helpful. I especially thought there was no point in trying to follow all the algebra and calculus. I would recommend it only for someone who is quite serious about learning about color photography, perhaps doing research, can follow the math, and spend the time on it.

Although the Evans book seems quite comprehensive, I have learned things easier from reading more general books and magazines on photography that were more straightforward and easier to follow, on topics such as color theory, color film structure, masking and processing, which I have an interest in. I can still use the Evans book as a reference book when questions arise, but I think the average starter is best served by more general publications.

 

[Ted Baker]

The IT people found the book long, tedious, and nearly incomprehensible.

Indeed

IMHO Kodak, Fuji et al, are missing a trick here, somebody or many somebodies holds the copyright for a number of defunct products, that many would be very interested in, myself included

Unfortunately these companies, don't have much experience in that side of things, or its not in the DNA to work that way.

 

[Photo Engineer]

Neither Evans.. nor Hunt cover digital. They talk about color and how it applies to film. They offer the math basis for that. Ed's book covers the gap to digital. So, all of them are useful. I would also add Mike Kriss to the mix. His work though mainly (the ones I have read) deal with image structure rather than with color.

For those uncomfortable with advanced math, these books might not be useful at all. But then, they would not be comfortable with the task at hand - writing a scanner program.

PE

 

[alanrockwood]

Neither Evans.. nor Hunt cover digital...

PE

The fifth edition of Hunt (1995) has a chapter on "Pictures from Computers" and another on "Digital Pictures and Desktop Publishing' but of course the information is somewhat limited, and it was before digital photography really took off.

 

[Photo Engineer]

Yeah, I had lunch with Bob it about 2008 and he was not really discussing digital in his books so far according to him. Here is a photo of our lunch group.

PE

 

[DREW WILEY]

Inkjet "inks" are so various and .complex in terms of individual gamut, transparency/opacity, and even gloss characteristics, that I doubt anyone even tries to really figure it all out. They just seem to read the result on what amounts to a fancy paint-matching reader which plots onto 4-dimensional CIE color-mapping space (in analytic geometry). The software takes over from there. A lot of distinct industries share similar technology. When my wife did DNA purifications in Biotech, she used a protein colorimeter made by the same company as made my darkroom densitometer. But her device cost 50K off the shelf, then another six million for the secret software. It was kept in a room with a timed bank vault door.

 

[Photo Engineer]

4D????? CIE is 3D at best.

I would send a file but it is too large.

PE

 

[DREW WILEY]

Yep. Industrial pigments are generally plotted 4d, with tone (whiteness) versus tint (blackness) on opposite axis. Quite a bit different from film. One reason is that black is never really black, but carries a particular hue bias, depending on the specifics. And some inkjet systems use several blacks which don't really match. One reason I dislike inkjet prints - too discontinuous.

 

[DREW WILEY]

More accurately, I should have stated, 4-axis and not 4D. That should clarify the confusion.

 

[Photo Engineer]

How do you draw 4 axes in 3D space? At best, the 3D Munsell system is quite useful but cumbersome.

PE

 

[Ted Baker]

Funnily enough, I suspect the real reason has little do with colour per se... even if the real reason turns out to be some historical convention. Sure colour is complex, but why is the bias in the negative and not in the paper? As an example It would be very possible to create negative/positive system using invisible dyes in the negative, as long as the paper could "see" them. My real interest centres around this conundrum with the negative/positive.

 

[Photo Engineer]

Bias in what way? Color or speed. In fact, color paper has a color bias in speed, but does not need one for color as the film is built to correct any paper errors.

A paper with invisible dyes was made once. You could only see the image with UV irradiation.

PE

 

[DREW WILEY]

Back to inks. Since neither white pigment is ever truly white, nor black ever black (for example, lampblack trend purplish upon dilution, and mineral blacks greenish), they split the light to dark axis at the center, and calculate each direction separately in analytic geometry. Most industrial spectrophotometers aren't continuous spectrum anyway, but flash at certain spacings of wavelengths (the more the better, but it's more practical than continuous tone). This just seems to have evolved to the point where it was very instinctive to adapt to the emerging inkjet applications, which are such complicated mixes of dyes, lakes, and micro-pigments that I don't see how else anyone would program those printers. No sense reinventing the wheel. Completely different from the basically transparent YMC mentality of film.

 

[DREW WILEY]

As per Munsell, I thought that was left in the dust decades ago. But none of these machines will ever be totally fool-proof from metamerism.

 

[Photo Engineer]

Currently, many printer inks used in production have metamerism. Photographic paper dyes have very little if any.

PE

 

[DREW WILEY]

The main problem with inkjet is that you can't choose ideal colorants. The priority is getting particles through those tiny nozzles. I do admire all the R&D that has gone into it, but I personally prefer the more seamless look of actual dyes.

 

[Ted Baker]

PE, follow up question

Am i correct that color negative film has different speed points, with blue layer typically the fastest?

Also do Kodak provide higher resolution of this graphs?