Photoshop Curves Preset for Color Negatives?

[msvsl]

Is there such a thing? Just a simple preset to load in curves, and voila...

Or is it just a dream?

 

[Bob Carnie]

Is there such a thing? Just a simple preset to load in curves, and voila...

Or is it just a dream?

yes you are dreaming... sliders on lightroom will work quite easily.

 

[StepheKoontz]

Repost of something I read about reversing color negs.

"The key step is pretty simple to do in Photoshop: sample the colour of the un-inverted negative rebate, make a new layer, fill the layer with the sampled colour, set blend mode to divide, flatten the layers, invert the image, clip RGB black & white points using warnings. Then fine colour adjustments & tonal balancing. The divide blending mode is essential - the mask is not a global colour - it's a mask that's formed inversely proportional to exposure & must be removed as such. If you do so, you're well on your way to manually matching how an optical print responds."

 

[Adrian Bacon]

Repost of something I read about reversing color negs.

"The key step is pretty simple to do in Photoshop: sample the colour of the un-inverted negative rebate, make a new layer, fill the layer with the sampled colour, set blend mode to divide, flatten the layers, invert the image, clip RGB black & white points using warnings. Then fine colour adjustments & tonal balancing. The divide blending mode is essential - the mask is not a global colour - it's a mask that's formed inversely proportional to exposure & must be removed as such. If you do so, you're well on your way to manually matching how an optical print responds."

if memory serves, I was part of that discussion. That can get results, but the divide blend mode isn’t doing what the original author intended.

 

[Adrian Bacon]

Is there such a thing? Just a simple preset to load in curves, and voila...

Or is it just a dream?

Its mostly a dream. Even though C-41 is standardized, the fact of the matter is each emulsion has a unique gamma for each color channel, which in digital land if you want a good job you need to take into account.

 

[jim10219]

Yeah, sure. But you have to create them yourself. Also, they're not really good for final edits. They're more suited for rough draft stuff.

I have several different PS curves presets saved on my computer. They're all named for the type of film and the scanning method used, so if I scan some 35mm Portra 160 with my DSLR, I have a preset for that. And if I scan some 4x5 Ektar 100 with my Epson 4990, I have a preset for that. If you change the film or the scanning method, the preset won't work nearly as well, so they have a very limited value. That's why you have to make your own presets and not rely on other downloading them from other people. Your tools and methods will be different, so what works for them, won't work very well for you.

I basically just use the presets to make a proof. So I'll scan in all of my negatives, place them into folders on my computer that have the same film/scanning method applied to them, and then I'll batch process the entire folder. From there, it will give me an idea of which photos are worth working further and spending some time on, and which ones are going to just be filed away and most likely forgotten. Then, when I find one I want to actually print or whatever, I'll reopen the original scan, upload the curve preset, but instead of applying it, I'll use that as a starting point for making my final tweaks. It's more of a time saver than anything. You never want to just apply any presets blindly to a photograph. Not if you actually care about it. Each photo is different and needs to be treated as such.

The best thing about making your own presets for the PS curves tool is that you get to learn how to use the curves tool. A lot of people hate it and find it too complicated. However, once you learn it, it's actually fairly quick and easy to use. It just has a steep learning curve at the beginning. After that, it's no more difficult and time consuming than the crop tool.

 

[DREW WILEY]

Buy a color enlarger and you won't have to worry about it.

 

[Adrian Bacon]

Buy a color enlarger and you won't have to worry about it.

Almost nobody has a darkroom, nor the space for one, hence the desire to try the digital route.

 

[DREW WILEY]

I'm quite aware of that. But there is a more appropriate hybrid section of this site for that purpose.

 

[Ariston]

Repost of something I read about reversing color negs.

"The key step is pretty simple to do in Photoshop: sample the colour of the un-inverted negative rebate, make a new layer, fill the layer with the sampled colour, set blend mode to divide, flatten the layers, invert the image, clip RGB black & white points using warnings. Then fine colour adjustments & tonal balancing. The divide blending mode is essential - the mask is not a global colour - it's a mask that's formed inversely proportional to exposure & must be removed as such. If you do so, you're well on your way to manually matching how an optical print responds."

The irony is that I don't understand anything after the statement, "pretty simple to do in Photoshop"!

 

[DREW WILEY]

Oooops ... It appears that I'm the one trespassing. This is a hybrid section. Sorry about that.

 

[Lachlan Young]

if memory serves, I was part of that discussion. That can get results, but the divide blend mode isn’t doing what the original author intended.

As the original author of the quoted statement, as I now understand it, the sample and divide can get pretty close to the effect of printing with a 3200K source and 50R correction - which is a global correction, not what I wrote earlier. I'm disinclined to re-linearise the sensitometric curves of the film in the digital side (as opposed to ensuring that the reproduction medium is adequately linear) because it may potentially destroy the characteristics of the film chosen in the first place. Increasingly I'm more interested in seeing if I can adequately model the effects of the paper curve etc.

If done with a bit of care, the methodology will run very close to a good RA4 sort of look, but it'll also make pretty clear whether or not you've crossed curves through poor film process control (though those are correctable too in the individual channels). In a moment of boredom, I combined the whole procedure into a 3D LUT for Portra 400 and an Imacon scanner & it works pretty reliably - but I'd suggest that it's not a universal solution - I'd tend to say that it'll need tailored to each film and image capture device. The more 'universal' systems like the Frontier etc seem to perform some fairly brutal 'corrections' to force fit the film into an understanding of colour controllable by semi-skilled operators. In theory, given the standardisations of C-41, it should be possible to design a universal inversion model that works better than that of the Fuji Frontier etc for a given light source and sensor.

 

[StepheKoontz]

As the original author of the quoted statement, as I now understand it, the sample and divide can get pretty close to the effect of printing with a 3200K source and 50R correction - which is a global correction, not what I wrote earlier. I'm disinclined to re-linearise the sensitometric curves of the film in the digital side (as opposed to ensuring that the reproduction medium is adequately linear) because it may potentially destroy the characteristics of the film chosen in the first place. Increasingly I'm more interested in seeing if I can adequately model the effects of the paper curve etc.

If done with a bit of care, the methodology will run very close to a good RA4 sort of look, but it'll also make pretty clear whether or not you've crossed curves through poor film process control (though those are correctable too in the individual channels). In a moment of boredom, I combined the whole procedure into a 3D LUT for Portra 400 and an Imacon scanner & it works pretty reliably - but I'd suggest that it's not a universal solution - I'd tend to say that it'll need tailored to each film and image capture device. The more 'universal' systems like the Frontier etc seem to perform some fairly brutal 'corrections' to force fit the film into an understanding of colour controllable by semi-skilled operators. In theory, given the standardisations of C-41, it should be possible to design a universal inversion model that works better than that of the Fuji Frontier etc for a given light source and sensor.

The irony is that I don't understand anything after the statement, "as I now understand it".

I do know the technique I quoted gives me pretty good results.

 

[Lachlan Young]

The irony is that I don't understand anything after the statement, "as I now understand it".

I do know the technique I quoted gives me pretty good results.

If it works, it works!

 

[Adrian Bacon]

As the original author of the quoted statement, as I now understand it, the sample and divide can get pretty close to the effect of printing with a 3200K source and 50R correction - which is a global correction, not what I wrote earlier. I'm disinclined to re-linearise the sensitometric curves of the film in the digital side (as opposed to ensuring that the reproduction medium is adequately linear) because it may potentially destroy the characteristics of the film chosen in the first place. Increasingly I'm more interested in seeing if I can adequately model the effects of the paper curve etc.

If done with a bit of care, the methodology will run very close to a good RA4 sort of look, but it'll also make pretty clear whether or not you've crossed curves through poor film process control (though those are correctable too in the individual channels). In a moment of boredom, I combined the whole procedure into a 3D LUT for Portra 400 and an Imacon scanner & it works pretty reliably - but I'd suggest that it's not a universal solution - I'd tend to say that it'll need tailored to each film and image capture device. The more 'universal' systems like the Frontier etc seem to perform some fairly brutal 'corrections' to force fit the film into an understanding of colour controllable by semi-skilled operators. In theory, given the standardisations of C-41, it should be possible to design a universal inversion model that works better than that of the Fuji Frontier etc for a given light source and sensor.

In a lot of ways this is where we diverge when it comes to post processing, as I am a proponent of linearizing the characteristic curves of each color channel on the digital side. Yes, it does run the risk of altering the resulting color, biasing it towards a more sensitometrically accurate look, but once you're there and in a digital color managed environment, assuming your chosen color space is large enough, you can then alter the look to suit the output medium, whether that be a C-Print, or an inkjet print, or simply for digital display/web usage. That's just me though. There's lots of ways to get there and none of them are necessarily wrong, just different.

 

[Bormental]

I have those old threads bookmarked. As I am new to color film, my problem is that I do not have a frame of reference i.e. how a particular emulsion is supposed to look like? I have the same photo scanned by my lab, Epson V600 with Epson software, Plustek 120 Pro with Silverfast and with my camera and inverted by hand in Affinity Photo. All four scans look different!

Instead of Photoshop curves, I want a "reference" 16-bit linear TIFF in AdobeRGB space of a standard color target shot on every Portra variant by Kodak. I can make my own curves from there.

 

[Adrian Bacon]

I have those old threads bookmarked. As I am new to color film, my problem is that I do not have a frame of reference i.e. how a particular emulsion is supposed to look like? I have the same photo scanned by my lab, Epson V600 with Epson software, Plustek 120 Pro with Silverfast and with my camera and inverted by hand in Affinity Photo. All four scans look different!

Instead of Photoshop curves, I want a "reference" 16-bit linear TIFF in AdobeRGB space of a standard color target shot on every Portra variant by Kodak. I can make my own curves from there.

you can invest in a Macbeth colorchecker chart and shoot a correctly exposed frame in 5500K light that is as full spectrum as you can get it and use that as a reference. The patches are a known value, so you’ll know what you’re trying to get to.

 

[Bormental]

@Adrian Bacon This is more or less what I am trying to do (I have an X-Rite kit). The problem with it is that if I "overfit" to such reference I lose the unique color profile of the emulsion and end up with the same result as my digital camera. So I only align grey patches, but in this case I don't know if my development is on target. Regarding that, you have answered my cyan question earlier, see I couldn't tell if cyan was part of the charm, or my temps dropped too much.

 

[Adrian Bacon]

@Adrian Bacon This is more or less what I am trying to do (I have an X-Rite kit). The problem with it is that if I "overfit" to such reference I lose the unique color profile of the emulsion and end up with the same result as my digital camera. So I only align grey patches, but in this case I don't know if my development is on target. Regarding that, you have answered my cyan question earlier, see I couldn't tell if cyan was part of the charm, or my temps dropped too much.

the way I’ve dealt with that is going through that exercise with all the emulsions I can reasonably get today (portra 160/400/800, Ektar 100, gold 200, ultramax 400, Fuji cn200, superia 400, 400H) then making a profile that is the average of those. That way I have a “digital” RA-4 paper that each emulsion will render against and where the emulsions deviates from the average, that will come through.

it doesn’t match what an actual RA-4 optical print looks like, but the only way to get that “reference” is to actually make an RA-4 optical print of a frame with a Macbeth chart then measure the patches with a color print densitometer, an exercise I just don’t think is worth it as the emulsion is still expected to look reasonably accurate on a print. Kodak’s and Fuji’s RA-4 paper is essentially an average of their emulsions so that each emulsion will look reasonably accurate, but differences will come through, just like how I’ve created my digital version.

all that being said, in terms of contrast, the control strip reference I’m using right now has the red channel at 0.39 contrast, the green channel at 0.45 contrast, and the blue channel at 0.57 contrast. If you know your red is low, it’s easy enough to go into the curve for the red channel and pull it up. Your biggest challenge if doing this in PS will be doing a “raw” white balance to center your mid tones without changing the contrast of each channel. That’s where color starts to get weird and why you’re seeing differences between all the scans. Using the levels or tone curve to center the midpoints changes the contrast above and below the midpoint, which in turn changes the color mix. You really want to keep the same contrast across the whole range for each channel and center the midpoints by applying a multiplier. That’s where the magic starts to happen.

 

[Bormental]

ugh... sometimes I feel there's no way around it, as doing any kind of color adjustments is visually hard as you're staring at a freshly inverted desaturated gamma-1 image, and I will have no choice but dust off my C++ skills.

and speaking of that, why did you feel the need to make separate gray card exposures for each stop, vs having a single card with a gray scale (like the one often used to visualize the zone values)?

 

[Adrian Bacon]

ugh... sometimes I feel there's no way around it, as doing any kind of color adjustments is visually hard as you're staring at a freshly inverted desaturated gamma-1 image, and I will have no choice but dust off my C++ skills.

and speaking of that, why did you feel the need to make separate gray card exposures for each stop, vs having a single card with a gray scale (like the one often used to visualize the zone values)?

couple reasons for the separate grey card exposures vs using single card. I get a bit more control over which exposures I use, and it gives me the opportunity to map out significantly more of the dynamic range if I feel that I need to. For example, it’s not uncommon for somebody to send in a roll of 400H for processing they they went and over exposed by 2-3 stops because that’s the popular thing to do if using some trendy hipster film lab that encourages that sort of behavior. It’s handy to have a map of what that emulsion is doing way up there. The other reason is it’s just simpler to have a collection of frames with various exposure levels of grey cards for the other tools I’ve developed (and not really said anything about) for mapping out and characterizing the various emulsions. I have a whole collection of tools that I’ve written for the express purpose of inspecting various aspects of the tonal and color response as it relates to my scanning environment, and it’s just simpler and faster to have a gray card set up and make the exposures at the exposure levels I want to look at then feed that processed film through the tool set. I’ve also done a similar thing for mapping out the color response.

this is part of the reason why I’ve been so reluctant to just make my processing tool publicly available. There’s enough variation between actual scanning hardware, and processed film that unless you have a really good understanding of how to make it work and why something is doing whatever, you’re at the same place you are now, except you have a tool that’s even more confusing to use because it’s working in raw color and manipulating things to arrive at a target on the other side of a colorspace transform engine in Adobe Lightroom. It’s a pretty tall order to support my own activities, having to provide support to an end user that doesn’t necessarily understand anything about how any of this actually works and just wants something that works is a whole other thing.

 

[Bormental]

You have a bigger opportunity on your hands than scanning for clients. Perhaps not a VC-fundable one, but big enough to seriously consider.

 

[Lachlan Young]

Having had a go at building 3D-LUT's for doing most of the heavy lifting in inverting linear scans of colour negs to something close to how they might print optically, I can see where Adrian's coming from - the problem is building an interface that is user resistant & has capacity for operator/ machine error at the film processing & scanning stages. Fuji's Frontier software was/ is probably the best attempt at building a turnkey system that didn't require a good prior understanding of colour correction etc, but it made quite significant compromises in gamut etc in order to be able to do so. For what it's worth, the Kodak Professional colour negative films are aimed to generally follow the same baseline corrections as each other to bring them to their nominal 'neutral' position relative to their warmth/ neutrality/ saturation.