How do scanners color correct C-41 negatives?

[Derek L]

Not a simple answer, but consider the following:

Thank you, this is very helpful.

Now back to the negative, since it not designed to viewed by the human eye there is no need for any specific color to represent the primaries, all you need is for those primaries to be "viewable by the print stock" you could for example have one dye that restricted UV light, one dye that restricted visible light and one dye that restricted infra red. Now if they could actually build such a system with widely separated dyes and print stock that could see these dyes, it would work perfectly and you would also not need any masking. You would not be able to scan the film with conventional DSLR though. The original technicolor actually just used three rolls of black and white film, so this concept has been used before.
The interaction between color negative and positive mimics part of this concept with the red sensitive layer. The Blue and Green layers of the print stock have similar spectral sensitivity to traditional camera negative, and are also close to spectral sensitivity of the Blue and Green layers of your DSLR. These are the layers that also have the masking, to reduce cross talk from adjacent dyes. The Red sensitive layer of the print stock is different from both negative film, a DSLR and anything designed to match the human eye in some way. It has for example a blind spot which allow the use of color safelight, right where the peak sensitivity of red sensel of your DSLR is and not far from the peak sensity of the cones in your eye sensitive to long wavelenght light. And the peak sensitivity of the red layer of the paper is where the infrared layer of your DSLR cuts our most of the light. Having the red sensitivity peak around 700nm reduces cross talk as the red layer cannot also have a mask. i.e. you can't have a three colour mask.

So every triplet from you DSLR is slightly wrong, This cannot be corrected just with just a 1D transformation, the 10bit cineon system can represent 107 million triplets, the 10bit log encoding was sized to fit dynamic range of color negative film. Now there is no need nor is it possible to measure the correction needed for all 107 million triplets, but you need more than 1D transformation using a curves to model the system.

OK. So let's iterate my proposed model once more. We DSLR scan the negative in a linear representation, then multiply the green and blue channels to render the film rebate black (i.e. account for the differing speeds PE pointed out). Now, you say the blue and green channels of the DSLR scan should be more less "right" at this point, since the spectral sensitivities are nearly the same as the paper. So I would expect to see at this point, if I shot a greyscale target, neutrals that have representation (r,b,g) where b=g (roughly) and r is some crazy number due to the different spectral response. At this point it remains to correct the red. Since b=g (roughly) and we know their values are correct (roughly), any 3x3 matrix transform is going to boil down to computing r_new = r_old + c*g for some coefficient c. This can be done with some coding, or just by appropriately adding the channels in photoshop. I can choose c to minimize the shift from perfect neutrals along a greyscale, say.

I suppose at this point I need to start empirically evaluating my ideas, and perhaps searching the academic literature on automatic color correction. But, what do you think of this? Obviously it's just a linear model, but presumably the nonlinearities of the system aren't so bad.

(The comment the book makes about the DSLR response following the Status M response is somewhat confusing and I'm deferring to your more precise answer here.)

 

[Adrian Bacon]

OK. So let's iterate my proposed model once more. We DSLR scan the negative in a linear representation, then multiply the green and blue channels to render the film rebate black (i.e. account for the differing speeds PE point out). Now, you say the blue and green channels of the DSLR scan should be more less "right" at this point, since the spectral sensitivities are nearly the same as the paper. So I would expect to see at this point, if I shot a greyscale target, neutrals that have representation (r,b,g) where b=g (roughly) and r is some crazy number due to the different spectral response. At this point it remains to correct the red. Since b=g (roughly) and we know their values are correct (roughly), any 3x3 matrix transform is going to boil down to computing r_new = r_old + c*g for some coefficient c. This can be done with some coding, or just by appropriately adding the channels in photoshop. I can choose c to minimize the shift from perfect neutrals along a greyscale, say.

I’ve been watching this thread with great interest... you also have to take into account the color of the light (and how much spectrum it’s emitting) that you’re using to illuminate the film with for scanning because it will effect things. You can see an example of this with color gels. Rosco publishes tech sheets for each of their color gels and spectral response with different light sources. The same gel doesn’t respond the same way under all lights. The same applies to film as it’s effectively a (very complex) color gel, and the DSLR used as it’s response is also effected by the color of the light used.

 

[MattKing]

I too am watching all these threads with considerable interest (and a fair bit of trepidation as well).
For DSLR scanning of C41 colour negatives, would I be correct in assuming that it will be advantageous to use a more continuous spectrum, halogen or incandescent light source to provide illumination, rather than a less continuous LED or fluorescent source?
EDIT: and how does the output from electronic flash fit into this?
If so, how should one approach using the white balance controls in the DSLR (assuming one wants jpeg output in addition to RAW)?

 

[Photo Engineer]

Alan, it is an Epson Perfection 4870 and I use Epson software for drivers and Photoshop for the interface.

PE

 

[Photo Engineer]

As to the above comments, we have built materials sensitive to UV, Visible and IR, and dyes that absorb in the same regions, but they are impractical. You can probably see the reasons why.

Ed. Giorgianni is an excellent scientist and I knew him when I worked at EK. His work is very advanced and could be misunderstood by some without the background. My own background is by doing, as that is all we had back when I designed films and papers. My teacher was Sir Robert Hunt of the Kodak Harrow labs.

PE

 

[Adrian Bacon]

I too am watching all these threads with considerable interest (and a fair bit of trepidation as well).
For DSLR scanning of C41 colour negatives, would I be correct in assuming that it will be advantageous to use a more continuous spectrum, halogen or incandescent light source to provide illumination, rather than a less continuous LED or fluorescent source?
EDIT: and how does the output from electronic flash fit into this?
If so, how should one approach using the white balance controls in the DSLR (assuming one wants jpeg output in addition to RAW)?

In my experience, more spectrum gives better results. I use a strobe as it’s pretty high CRI, and gives me enough light output (in a controllable manner) that I can shoot with the lens aperture closed down enough to get good depth of field and still have a fast shutter speed.

 

[MattKing]

In my experience, more spectrum gives better results. I use a strobe as it’s pretty high CRI, and gives me enough light output (in a controllable manner) that I can shoot with the lens aperture closed down enough to get good depth of field and still have a fast shutter speed.

Does the CRI value for a light source give a reliable indication of the source's suitability for these purposes? I ask because it seems to me that a tungsten light bulb may have a really low CRI due to its colour, but may be really useful for printing colour negatives. In contrast, some LED sources have high CRIs, but I'm not sure that they are immune from exhibiting some important discontinuities in their spectrum. Or does having a high CRI mean that the spectral discontinuities are not important?

 

[MattKing]

More generally, and maybe this question reflects an important misunderstanding on my part, I don't understand why so much attention is being paid to the mask.
My understanding is that the film plus the mask work together to solve all the issues and to make the image forming light suitable for the paper.
My question is: Why not simply design software settings to respond the same way as the paper does (including all necessary inversions)?

 

[Mr Bill]

it seems to me that a tungsten light bulb may have a really low CRI due to its colourful...

Actually a tungsten bulb will have a very high CRI, probably on the order of 99 or 100 (100 is "perfect). The general method, as I recall, is to reference against a black body radiator of the same color temperature. (Best to double check the method before quoting me, though; I'm just going by memory.)

 

[Adrian Bacon]

Does the CRI value for a light source give a reliable indication of the source's suitability for these purposes? I ask because it seems to me that a tungsten light bulb may have a really low CRI due to its colour, but may be really useful for printing colour negatives. In contrast, some LED sources have high CRIs, but I'm not sure that they are immune from exhibiting some important discontinuities in their spectrum. Or does having a high CRI mean that the spectral discontinuities are not important?

This is useful: https://en.wikipedia.org/wiki/Color_rendering_index

My question is: Why not simply design software settings to respond the same way as the paper does (including all necessary inversions)?

Because the scanner (DSLR or otherwise) hardware doesn't respond the same way as paper. You have to characterize the hardware's response to the film, and once you've done that, since you're digitizing it, you might as well then make it fit to a digital color space instead. If you do that accurately, you can output it to any other color space, including printers that have ICC profiles, or any other output device that has an ICC profile. Where most people tend to go wrong is they try to characterize the RA4 paper and get the scan to fit into that, then wonder why it doesn't look right on screen. Every other digital output device operates inside of ICC profiles, if you don't first conform the scan to that, it will never look right on a display that operates by that standard. It's like trying to make two different systems work together. Why? Forget one, and just use one system. In the case of digital, that would be ICC color spaces. RA4 paper is an analog color system. You could make an ICC profile that characterizes it, but it's the same amount of work to fit the scanned film to that as it would be for any other color space, so you're better off going with a pretty large industry standard color space. That used to be Adobe RGB, but currently ProPhoto is more prevalent, and the ACES family of color spaces is the future.

 

[Ted Baker]

(The comment the book makes about the DSLR response following the Status M response is somewhat confusing and I'm deferring to your more precise answer here.)

Status M is a much better match to the paper than the DSLR. So the DSLR is moving in the wrong direction...

OK. So let's iterate my proposed model once more. We DSLR scan the negative in a linear representation, then multiply the green and blue channels to render the film rebate black (i.e. account for the differing speeds PE pointed out). Now, you say the blue and green channels of the DSLR scan should be more less "right" at this point, since the spectral sensitivities are nearly the same as the paper. So I would expect to see at this point, if I shot a greyscale target, neutrals that have representation (r,b,g) where b=g (roughly) and r is some crazy number due to the different spectral response. At this point it remains to correct the red. Since b=g (roughly) and we know their values are correct (roughly), any 3x3 matrix transform is going to boil down to computing r_new = r_old + c*g for some coefficient c. This can be done with some coding, or just by appropriately adding the channels in photoshop. I can choose c to minimize the shift from perfect neutrals along a greyscale, say.

You starting to see and solve part of the problem yes. But you also have to remember the image on the negative is encoded using a gamma less than 1. It is about .6, in the mid tones and the paper has a gamma of about 3 or 4 to give a final midscale gamma of around 1.8, roughly. In order to matrix calcs like you describe you will need to linearise the data encoded in the negative to 1.0 before applying the transformation. This will give you a internal model that represents the negative, you then need to model the response of the paper.

Read how cineon works and read about Kodak Photo CD... Your on the right track.

This thread describes some of the tools that I wrote. The clever stuff with colour I did not release, because I am considering releasing a commercial product, but It may be of interest. https://www.photrio.com/forum/threads/scantools-a-few-tools-to-help-with-film-scanning.161917/

 

[Ted Baker]

My question is: Why not simply design software settings to respond the same way as the paper does (including all necessary inversions)?

Modern digital software all works with linear models, 1,2,3,4,5,7,8,9,10 etc. The maths to do transformations works best this way The camera sensors are essentially linear, any photo that you see as finished output, including RAW photos that have been processed in something like light room have had a tone curve applied. This is fundamentality different to analogue photography where at the capture stage (ie. exposure) the film has a tone cure AKA characteristic curve built in.

If Kodak still sold zillions of rolls of color negative, we would indeed have affordable and up to date software like lightroom and hardware designed for the job. Instead we have to make do. Which is what some of the developers who have posted here are trying to do.

For DSLR scanning of C41 colour negatives, would I be correct in assuming that it will be advantageous to use a more continuous spectrum, halogen or incandescent

That is probably the safe bet, as since the sensor is mismatched using a broad spectrum, ensures that the mismatch is not increased. If you use three narrow spectrum lights you have to make sure you choose them correctly as you have a more than 50% chance of making things worse.,..

 

[MattKing]

I appreciate all your efforts.
Correct me if I am wrong, but any attempt to deal with a particular film's mask is unnecessary and not well founded, because it is the combination of the mask (which varies with the image) and the image itself that determines the matter.
I still wonder whether or not there might be an advantage to basing any of these approaches on the one standard that I believe is in place - the response of the colour paper.
It seems to me that that response is essentially a colour space of its own - even though it is a non-linear response - and that all "normal" C41 films are designed to output to that colour space.
So, if one determines the transformation necessary for one sensor to work in that colour space, than one has determined the transformation necessary for all films.
I understand that then one must further transform the result in order to view it, or print it, but isn't that what is done anyways?
And of course if a transformation function is sensor specific (or should I say sensor plus processor plus firmware specific?) would it not be easier to develop sensor to sensor transformations?

 

[Adrian Bacon]

I understand that then one must further transform the result in order to view it, or print it, but isn't that what is done anyways?

Yes, but in a color managed environment, all that is done automatically under the covers. The challenge is getting the raw scanner samples conformed to an ICC color space in the first place. Once you're there, the rest just happens for you by the color managed environment.

 

[runswithsizzers]

I recently exposed a couple of rolls of Ektar. One I scanned myself (both with a film scanner and by copying the negs with my dSLR) - and the other roll I had "scanned" at the time of processing (by member Adrian Bacon). It was fun, and I was pleased, overall, with the results. Examples <here> and <here>

However. After reading this thread and a similar one: "An alternative to Negative Lab Pro and Lr has to exist (C-41 reversal and orange mask removal)?!" I think I am going to forget about trying to digitize color negative film for a while, and go back to shooting slide film. I don't know that my results will be any better, but I need to give my brain a rest from trying to understand how to best d.i.y. color correct scans/dSLR copies of C-41 negatives.

 

[Alan Edward Klein]

Doesn't Hollywood film with negative film and then digitized by scanning it. How do they get rid of the mask?

 

[Alan Edward Klein]

I recently exposed a couple of rolls of Ektar. One I scanned myself (both with a film scanner and by copying the negs with my dSLR) - and the other roll I had "scanned" at the time of processing (by member Adrian Bacon). It was fun, and I was pleased, overall, with the results. Examples <here> and <here>

However. After reading this thread and a similar one: "An alternative to Negative Lab Pro and Lr has to exist (C-41 reversal and orange mask removal)?!" I think I am going to forget about trying to digitize color negative film for a while, and go back to shooting slide film. I don't know that my results will be any better, but I need to give my brain a rest from trying to understand how to best d.i.y. color correct scans/dSLR copies of C-41 negatives.

This is why I shoot landscapes with Velvia in medium format. It's just easier to scan, you know what you got just looking at the film (I bracket), and it's easier and faster to post process to get the colors right. I don;t print chemically from the negative.

 

[Adrian Bacon]

I recently exposed a couple of rolls of Ektar. One I scanned myself (both with a film scanner and by copying the negs with my dSLR) - and the other roll I had "scanned" at the time of processing (by member Adrian Bacon). It was fun, and I was pleased, overall, with the results. Examples <here> and <here>

However. After reading this thread and a similar one: "An alternative to Negative Lab Pro and Lr has to exist (C-41 reversal and orange mask removal)?!" I think I am going to forget about trying to digitize color negative film for a while, and go back to shooting slide film. I don't know that my results will be any better, but I need to give my brain a rest from trying to understand how to best d.i.y. color correct scans/dSLR copies of C-41 negatives.

Hey! I recognize those photos in the second link... oh wait...

 

[Photo Engineer]

Status M is a much better match to the paper than the DSLR. So the DSLR is moving in the wrong direction...



You starting to see and solve part of the problem yes. But you also have to remember the image on the negative is encoded using a gamma less than 1. It is about .6, in the mid tones and the paper has a gamma of about 3 or 4 to give a final midscale gamma of around 1.8, roughly. In order to matrix calcs like you describe you will need to linearise the data encoded in the negative to 1.0 before applying the transformation. This will give you a internal model that represents the negative, you then need to model the response of the paper.

Read how cineon works and read about Kodak Photo CD... Your on the right track.

This thread describes some of the tools that I wrote. The clever stuff with colour I did not release, because I am considering releasing a commercial product, but It may be of interest. https://www.photrio.com/forum/threads/scantools-a-few-tools-to-help-with-film-scanning.161917/

Ted, negative MP films are about gamma 0.55, professional still about 0.60 - 0.65 and amateur/consumer are about 0.75+ having been adjusted for the (assumed) higher flare of cheaper cameras.

Color paper is about 2.5 which gives 0.6 x 2.5 or about 1.5.

PE

 

[MattKing]

Doesn't Hollywood film with negative film and then digitized by scanning it. How do they get rid of the mask?

You don't want to get rid of the mask. The mask corrects the inaccuracies of the emulsions.
The "sandwich" of mask and emulsions is a corrected one which, nicely matches:
1) for still film, the paper emulsion, which does a fine job of filtering out the mask's colour cast, inverting the result and presenting it for print display; and
2) for motion picture film, the print stock emulsion, which also does a fine job of filtering out the mask's colour cast, inverting the result and presenting it for projection display.
In the case of modern motion pictures, they scan the negatives, invert the results, edit the inverted result and then, if the result is going to be optically projected, re-invert and re-mask it, for printing to projection stock.

 

[Alan Edward Klein]

You don't want to get rid of the mask. The mask corrects the inaccuracies of the emulsions.
The "sandwich" of mask and emulsions is a corrected one which, nicely matches:
1) for still film, the paper emulsion, which does a fine job of filtering out the mask's colour cast, inverting the result and presenting it for print display; and
2) for motion picture film, the print stock emulsion, which also does a fine job of filtering out the mask's colour cast, inverting the result and presenting it for projection display.
In the case of modern motion pictures, they scan the negatives, invert the results, edit the inverted result and then, if the result is going to be optically projected, re-invert and re-mask it, for printing to projection stock.

No wonder the average photographer has problems with it.

 

[jtk]

No wonder the average photographer has problems with it.

I think that "average photographer" has no problems if he/she simply follows instructions and uses proper PS or Elements. Where people have troubles comes from trying to avoid direct, easy answers.

One direct, easy answer...perhaps the easiest..is NIK.

 

[Alan Edward Klein]

I think that "average photographer" has no problems if he/she simply follows instructions and uses proper PS or Elements. Where people have troubles comes from trying to avoid direct, easy answers.

One direct, easy answer...perhaps the easiest..is NIK.

Clarify.

 

[Les Sarile]

Been meaning to try a Nikon D850 as it is full frame (perfect for my Nikon autobellows) and it has a built-in color negative conversion. It would be great if it is as good as the Nikonscan+Coolscan color conversion which is the best I've found to date.

 

[runswithsizzers]

I think that "average photographer" has no problems if he/she simply follows instructions and uses proper PS or Elements. Where people have troubles comes from trying to avoid direct, easy answers.

One direct, easy answer...perhaps the easiest..is NIK.

I have the old (free) NIK collection. I have used the Silver Efex Pro 2 module for B&W effects - but I have not really explored the other modules. Does the NIK collection include a function to invert and/or color correct C-41 negatives?