Why the difficulty of producing an acceptable positive from C41 film in the digital conversion?

[Steve@f8]

After digitising a B&W negative, the process of producing a positive in digital post-processing is a relative doddle, but doing the same from a colour negative is a massive challenge. Why is this? I understand the orange mask has to be negated, but after that it would seem (before actually trying to do it yourself) to be a ‘simple’ task of inversion and adjusting the sliders to taste, colours, contrast, brightness etc. NLP and other software packages must have some clever algorithms to do the conversion by analysing the digitised negative. I use LR 6.14 and have almost given up post processing colour negatives with tools I have available to me.

 

[loccdor]

You didn't mention if you're doing this digitization with a camera or a scanner.

I do it with a camera mainly, and one of the things I've noticed that orange-masked film is particularly affected by is stray light from not masking your backlight source perfectly. With slide film and a white light source, stray light is going to have the same color temperature as the light going through your film, and the more contrasty brightness range will mitigate it further. But with orange mask C-41, any stray light which is not being filtered through the orange mask is going to result in a different color temperature flare which may be near-impossible to mitigate in post processing. C-41 also has less contrast within each channel which means the effect of a flare will be greater since you are magnifying the contrast.

Assuming stray light is not the issue (you've masked and hooded things well, or are using a scanner), then if you are digitizing C-41 in 16-bits-per-channel it's important to look at the individual R-G-B histogram and make sure none of the channels are getting truncated. That will give you a RAW or TIF that you can work well with.

Having surmounted those two C-41 obstacles, you can invert, and then set your black point to the minimum of each of the R-G-B channels individually, and your white point to the maximum of the channels, in order to normalize the image and stretch the values over the full range available.

Then, you can pull the R, G, B curves up or down individually until you stop seeing unnatural color casts (unless desired for the effect). This is the part that takes a while if done manually, as I do in Gimp. Sometimes 5-10 minutes to get it right in an image I really care about. It does take practice to get the feel for what colors look natural as the eye adjusts. Usually pulling from the middle of the curves with a single point is sufficient, except in complicated cases. When you work on a whole roll of a particular type of film, you tend to see that you're making similar color adjustments on every image, unless some are badly under or over exposed. So it gets faster as you practice.

Of course, you can get more advanced than me and use software specifically designed to handle color negative, which may streamline some of these steps in some cases, at the loss of some manual control.

 

[xkaes]

Steve, we all feel your pain. One would think that a simple solution would have been created a long time ago. Film scanners do it instantaneously, so why isn't there software available to do the same damn thing?

There is, and it's been discussed a lot on this FORUM -- do a SEARCH for previous threads (upper right hand corner of this screen).

That being said, the software available has its PROS & CONS, from cost to complexity (see Post #2) to accuracy.

Perhaps your resurrecting this discussion will uncover some recent advances -- if not perfect solutions.

 

[bags27]

I have 2 programs--each costing ~ $100: NLP and NegMaster. $100 is about 7 rolls of Portra 400 or about 2 batches of C-41 developer, but is inexhaustible.

I preferred NegMaster to the previous version of NLP. But NLP improved and, for some reason, I can't get NegMaster to work on my most recent operating system.

NLP is just a bit of a drag because it works only in LR, meaning I develop there and then send the photos to Photoshop. Small potatoes, really.

At that point, I'm the artistic director of my own work. Film is a (beautiful) representation (not reproduction) of reality. At every step--from choosing the film, lens, exposure, etc. there are variables. I think of the final work in Photoshop is just another variable I'm responsible for.

 

[koraks]

it would seem (before actually trying to do it yourself) to be a ‘simple’ task of inversion and adjusting the sliders to taste, colours, contrast, brightness

As far as I'm concerned, that's pretty much what it boils down to, yes. I generally don't touch contrast or brightness sliders and only do curve adjustments.

Flipped – doing color negative inversions manually – Koraks tinkers

tinker.koraks.nl

I find it easier if I scan several frames in one go (flatbed) and do an inversion + adjustment curve on all of them at once. Then scan all frames of the same film with the same settings (without any autocorrection enabled in the scanning software) and apply the same curve adjustment to the whole batch. For me, this is the quickest and most consistent way of doing it. YMMV and all that.

 

[Steve@f8]

As far as I'm concerned, that's pretty much what it boils down to, yes. I generally don't touch contrast or brightness sliders and only do curve adjustments.

Flipped – doing color negative inversions manually – Koraks tinkers

tinker.koraks.nl

I find it easier if I scan several frames in one go (flatbed) and do an inversion + adjustment curve on all of them at once. Then scan all frames of the same film with the same settings (without any autocorrection enabled in the scanning software) and apply the same curve adjustment to the whole batch. For me, this is the quickest and most consistent way of doing it. YMMV and all that.

Thank you for the info and link, a good reference I’m sure I’ll use.
One thing in particular I picked up on is the way you deal with the dreaded orange mask. It doesn’t appear that you’re adjusting the white balance by using the dropper sampler on the film base, or on a neutral tone (if there is one, and if so can be recognised under the orange mask). instead you’re looking for a secondary peak in the RGB channels, predominantly red. Is this because of GIMP or for preference or better outcome?
Your processed versions are looking very good, and I would be pleased with similar results in my work flow

 

[Osmdesat]

While orange mask can be always compensated in post-process, I think much better is to pre-balance the negative image already using lightsource of appropriate colour or to use appropriate filter (I mean optical filter you put on the lens, not "filter" in Instagram app or whatever).
This way all the color channels will be exposed equally. If you use normal white light as backlight, the green and blue channels will always come out underexposed and you will have to amplify them in picture editor to balance the colors. So for me the starting point would be to have negative pre-balanced optically. Further digital balancing leads to loss of quality.
You can experiment with blue filter on the lens or very cold white LEDs like 9000K or so, however this will not be ideal.
Better to have gelatine CMY filters for color enlarging. you can pre-filter the backlight with suitable combination of theirs. If you have color enlarger with dichroic filter head, you can use this as a light source. Depends on your possibilities.
The desired compensating light will gonna have blueish-greenish tint.
I personally use diy RGB variable backlight for these purposes, but given that color LEDs produce narrow-band light, the colors in resulting scan will be somewhat off and they will have to be pre-processed for correct colorspace.

 

[Steve@f8]

You didn't mention if you're doing this digitization with a camera or a scanner.

I do it with a camera mainly, and one of the things I've noticed that orange-masked film is particularly affected by is stray light from not masking your backlight source perfectly. With slide film and a white light source, stray light is going to have the same color temperature as the light going through your film, and the more contrasty brightness range will mitigate it further. But with orange mask C-41, any stray light which is not being filtered through the orange mask is going to result in a different color temperature flare which may be near-impossible to mitigate in post processing. C-41 also has less contrast within each channel which means the effect of a flare will be greater since you are magnifying the contrast.

Assuming stray light is not the issue (you've masked and hooded things well, or are using a scanner), then if you are digitizing C-41 in 16-bits-per-channel it's important to look at the individual R-G-B histogram and make sure none of the channels are getting truncated. That will give you a RAW or TIF that you can work well with.

Having surmounted those two C-41 obstacles, you can invert, and then set your black point to the minimum of each of the R-G-B channels individually, and your white point to the maximum of the channels, in order to normalize the image and stretch the values over the full range available.

Then, you can pull the R, G, B curves up or down individually until you stop seeing unnatural color casts (unless desired for the effect). This is the part that takes a while if done manually, as I do in Gimp. Sometimes 5-10 minutes to get it right in an image I really care about. It does take practice to get the feel for what colors look natural as the eye adjusts. Usually pulling from the middle of the curves with a single point is sufficient, except in complicated cases. When you work on a whole roll of a particular type of film, you tend to see that you're making similar color adjustments on every image, unless some are badly under or over exposed. So it gets faster as you practice.

Of course, you can get more advanced than me and use software specifically designed to handle color negative, which may streamline some of these steps in some cases, at the loss of some manual control.

Yes, I forgot to mention I’m using a camera. (I also have a Plustek running Silverfast turning out good scans, but that has nothing to do with this.)
Stray light is something I’ve neglected in camera ‘scanning’, having huge amounts because I’m using a DigiaLisa for LC-Wide Panas, so there’s massive amounts top and bottom of the resulting frame captured. I also neglected to manually set the colour temperature in the camera, instead allowing AWB. I can rectify both of these and try again (I’ll have to guess the light panel output to 5000K or 5500K). I’ve noted the steps you’ve mentioned, eg the RGB channel adjustments and I’ll work in a similar manner (sounds similar to Korak’s workflow in this thread).

 

[Steve@f8]

While orange mask can be always compensated in post-process, I think much better is to pre-balance the negative image already using lightsource of appropriate colour or to use appropriate filter (I mean optical filter you put on the lens, not "filter" in Instagram app or whatever).
This way all the color channels will be exposed equally. If you use normal white light as backlight, the green and blue channels will always come out underexposed and you will have to amplify them in picture editor to balance the colors. So for me the starting point would be to have negative pre-balanced optically. Further digital balancing leads to loss of quality.
You can experiment with blue filter on the lens or very cold white LEDs like 9000K or so, however this will not be ideal.
Better to have gelatine CMY filters for color enlarging. you can pre-filter the backlight with suitable combination of theirs. If you have color enlarger with dichroic filter head, you can use this as a light source. Depends on your possibilities.
The desired compensating light will gonna have blueish-greenish tint.
I personally use diy RGB variable backlight for these purposes, but given that color LEDs produce narrow-band light, the colors in resulting scan will be somewhat off and they will have to be pre-processed for correct colorspace.

I only have a bog standard light panel and I guess it’s somewhere in the region 5000 to 5500K. No filters at my disposal either, but your observations and tips are really helpful. Thank you.

 

[Steve@f8]

Steve, we all feel your pain. One would think that a simple solution would have been created a long time ago. Film scanners do it instantaneously, so why isn't there software available to do the same damn thing?

There is, and it's been discussed a lot on this FORUM -- do a SEARCH for previous threads (upper right hand corner of this screen).

That being said, the software available has its PROS & CONS, from cost to complexity (see Post #2) to accuracy.

Perhaps your resurrecting this discussion will uncover some recent advances -- if not perfect solutions.

Thank you, I’ll do a search which I guess will be in the hybrid section.
You’re absolutely spot on, this process is a right royal pain in the back end.

 

[xkaes]

Years ago I read an article -- Darkroom Techniques? -- in which the author discussed using the filters in a colorhead to mimic the orange mask. The article was unrelated to digital photography, but it does seem possible to cancel out the orange mask(s) by using a color head -- or as you sorta suggested doing a custom white balance with a blank film exposure.

 

[Osmdesat]

I only have a bog standard light panel and I guess it’s somewhere in the region 5000 to 5500K. No filters at my disposal either, but your observations and tips are really helpful. Thank you.

You can experiment with decorative color transparent foils found in paper shops. Search for blue-green or cyan tints. Then put the foil onto the backlight. The foils however are not optical grade, so the negative must be little distant from the surface. I did some experiments with coloured PET bottles from various soda drinks. I cut out desired shape from the cylindrical part of the bottle.

 

[Alan Edward Klein]

I switched to chromes.

 

[koraks]

One thing in particular I picked up on is the way you deal with the dreaded orange mask. It doesn’t appear that you’re adjusting the white balance by using the dropper sampler on the film base, or on a neutral tone (if there is one, and if so can be recognised under the orange mask). instead you’re looking for a secondary peak in the RGB channels, predominantly red. Is this because of GIMP or for preference or better outcome?

The pipette method would give roughly the same result the way I do it; in either case, you 'look for' (or have PS/GIMP etc. do it) the color of the orange mask and define that as the black point. Whether that's done with curve adjustment, a pipette or some other method isn't very important. I do it with curves because one single adjustment in just the curves window gets me pretty much where I need to be. A multi-step process is of course just as good, as long as the same steps are replicated for all frames (provided you want consistency across frames).