Creating custom camera profiles for digital camera "scanning" of slides using an IT8 target: anyone tried this?

GLS · Mar 6, 2023

My current setup for digitising film is a Sony A7RIV with Sigma 70mm ART macro lens, and a Kaiser Slimlite Plano LED panel as light source. RAW capture is used.

Whilst this works, when digitising slides the colours and contrast obtained "out of the gate" are significantly different to the appearance of the original to the eye whilst on the lightbox, requiring quite a bit of tweaking of HSL sliders, RGB curves and so on to match things by eye. Whilst I have become reasonably adept at this, it's hard if not impossible to match it exactly, which can be frustrating. Ideally I would like a better way that is more precise and reproducible, achieving something that is at least very close to the original from the get-go, requiring minimal further tweaking.

Clearly the answer is a custom profile. From my understanding, in theory this would involve purchasing a hand-measured transmissive IT8 target (such as those from LaserSoft), photographing it with my setup above (at a fixed exposure), then using that RAW data to create a DCP or ICC profile with an appropriate profiling program, such as the open-source DCamProf, or its commercial GUI version Lumariver Profile Designer.

Has anyone here done something like this, and could offer advice or suggestions? How easy is it to do, are there any pitfalls etc?

Many thanks in advance.

L Gebhardt · Mar 8, 2023

My suspicion is your LED panel may be the issue. Before you go to all the trouble of profiling try exposing with actual daylight, or at least an incandescent bulb or flash. My experience is slides come out pretty close when I use a flash or a color enlarger head adjusted to be close to daylight balance.

GLS · Mar 8, 2023

The panel has a CRI of 95. Not perfect, but not too shabby either. My longer term plan is to eventually switch to a 99+ CRI continuous light source (flash is no good for my workflow) for ultimate colour accuracy, but I don't think that is the major factor at play here. Rather my take on it is simply that none of the default Adobe or Sony RAW profiles do a great job of accurately reproducing the hues of the dyes used in films (either slide or colour negs), because they were not designed for that. The "neutral" variants get closest on average, but still leave much to be desired IMO.

MattKing · Mar 8, 2023

If you decide to go down that particular rabbit hole, you will discover that "CRI" is a far from ideal measurement standard for the things that matter for colour digital photography.
There are other more appropriate - and more complex - standards, but you won't find the results listed in respect to most light sources easily available to the layperson.
The motion picture lighting industry uses some of them.

Adrian Bacon · Mar 11, 2023

A simple way to do this is to take a photograph (on your slide film of choice) of an x-rite color checker card (using daylight, or flash), then digitize it with raw, then use that raw digitized frame to make a custom camera profile in Lightroom. There's no need to use an IT-8 target, those were for back in the day when people used scanners, not cameras. You can google how to make a profile with the color checker card. It's pretty straightforward.

GLS · Mar 11, 2023

Adrian Bacon said:
A simple way to do this is to take a photograph (on your slide film of choice) of an x-rite color checker card (using daylight, or flash), then digitize it with raw, then use that raw digitized frame to make a custom camera profile in Lightroom. There's no need to use an IT-8 target, those were for back in the day when people used scanners, not cameras. You can google how to make a profile with the color checker card. It's pretty straightforward.

Thanks for the reply. I understand your point, but isn't that in effect just trying to make your own simplified version of an IT-8? If ultimate accuracy in reproduction of the slides is the goal, isn't my proposed method better? After all, an IT-8 has more patches to measure off and therefore will need less interpolation in the profile. A HutchColor HCT target would be even better, but those are insanely expensive.

Adrian Bacon · Mar 11, 2023

GLS said:
Thanks for the reply. I understand your point, but isn't that in effect just trying to make your own simplified version of an IT-8? If ultimate accuracy in reproduction of the slides is the goal, isn't my proposed method better? After all, an IT-8 has more patches to measure off and therefore will need less interpolation in the profile. A HutchColor HCT target would be even better, but those are insanely expensive.

I wouldn’t call it simplified. The x-rite color checker puts its patches in key color locations that are specific to how digital cameras see, and those patches are also the brightness and saturation that are specifically for digital cameras.

sure, you can add more patches, and be more accurate on paper, but perceptually, not so much. Also, in my own experience of making scanning profiles for film, more patches and more accuracy doesn’t usually look better, as it tends to introduce perceptual discontinuities in hue and saturation rendering. Less patches and a little more smooth interpolation tends to be more pleasing to the eye.

GLS · Mar 12, 2023

Adrian Bacon said:
I wouldn’t call it simplified. The x-rite color checker puts its patches in key color locations that are specific to how digital cameras see, and those patches are also the brightness and saturation that are specifically for digital cameras.

sure, you can add more patches, and be more accurate on paper, but perceptually, not so much. Also, in my own experience of making scanning profiles for film, more patches and more accuracy doesn’t usually look better, as it tends to introduce perceptual discontinuities in hue and saturation rendering. Less patches and a little more smooth interpolation tends to be more pleasing to the eye.

I see. Yes, I suppose there will always be a trade-off between colourimetric and perceptual accuracy. Hmmm, something to think about then.

I guess just the one such profile from the checker taken in direct sunlight should be used then, rather than different ones taken under different lighting conditions? Otherwise I assume you would be "correcting" out the white balance of the scene?

Steven Lee · Mar 12, 2023

@GLS I am in the middle of doing exactly this. The problem is with obtaining a high quality transmission target. They fade over time and I do not trust those old Kodak targets with unknown storage history sold on eBay. I recently purchased Wolf Faust targets, but the first profile I generated made things slightly worse (magenta cast) and haven't found the time to try again.

You can also reach out to André Eikmeyer of NEGMASTER fame and he can make you a profile specifically for your camera+light source combination.

I am not sure about Adrian's suggestion of using a reflective target, I am not an expert in color management. Based on my limited understanding, it matters which colors are used for the patches and it matters whether the light source used for profile generation is the same as the one for "scanning". Spectral sensitivity peaks of camera sensors are optimized for perceptual accuracy of reflected light and they are supplemented with built-in color profiles for two light sources: daylight + tungsten. The peaks on film are different and all of us have different light sources.

GLS · Mar 12, 2023

Steven Lee said:
The problem is with obtaining a high quality transmission target.

I ordered a new 4x5 Fujichrome one from LaserSoft. Still waiting for it to arrive though.

Adrian Bacon · Mar 12, 2023

GLS said:
I see. Yes, I suppose there will always be a trade-off between colourimetric and perceptual accuracy. Hmmm, something to think about then.

I guess just the one such profile from the checker taken in direct sunlight should be used then, rather than different ones taken under different lighting conditions? Otherwise I assume you would be "correcting" out the white balance of the scene?

Ideally, you’d expose the colorchecker on film using the native color balance of film, I.e. 5500K. You could use direct sunlight, but I personally prefer strobe At full power so you get the full strobe output spectrum. if possible, also expose the white balance panel of the card with the same light (This is with the x-rite color checker passport)

once it’s processed, the frame with the white balance panel should be used to set the white balance of your camera and light source as seen through the film. From there, you can digitize the other frame with the patches and run it through the profiler tool.

this will level out both the rendering of the slide and camera, which will be better than your previous starting point, but probably not what you want In terms of saturation. The better thing to do would be to use the x-rite card to profile just the camera, ideally using the same light source you’re using to digitize the film, this way the camera itself is reasonably accurate in terms of hue, brightness and saturation, so when you digitize the slide film, all the saturation wil pop through. Make sure you use a color space that is big enough, and ideally your monitor should cover at least 100% of sRGB and most if not all of Adobe RGB so you can actually see the slides in most of their glory.

wiltw · Mar 12, 2023

The light source certainly can be a contributor to color reproduction issues. But let me state that even when shot with Daylight full spectrum lighting, conversion software for neg-to-positive image conversion can be a real pitfall of the color reproduction and other issues! See this recent Photrio post of mine...

Is DSLR scanning "better" than flatbed scanning?

It should be easy enough to make one that does 6x6 in 4 shots. A square neg carrier, when slid up against the top left corner of the light box would have the lower right portion of the neg in position to be scanned. Sliding the carrier to the top right puts lower left part of the neg in...

www.photrio.com

GLS · Mar 12, 2023

wiltw said:
The light source certainly can be a contributor to color reproduction issues. But let me state that even when shot with Daylight full spectrum lighting, conversion software for neg-to-positive image conversion can be a real pitfall of the color reproduction and other issues! See this recent Photrio post of mine...

Is DSLR scanning "better" than flatbed scanning?

It should be easy enough to make one that does 6x6 in 4 shots. A square neg carrier, when slid up against the top left corner of the light box would have the lower right portion of the neg in position to be scanned. Sliding the carrier to the top right puts lower left part of the neg in...

www.photrio.com

I'm aware of the added complications with inverting colour negs, but this thread is specifically about reproduction of slides.

GLS · Mar 17, 2023

So an update on this.

I decided to splurge some more and do this thing properly. Bought the Negative Supply Scanning Light Source Pro 5x7" 99 CRI. Overpriced, but extremely well made and ideal for my existing setup; I discovered by happy chance that my custom-made 4x5" film scanning holder sits perfectly astride the built in plastic mask that the NS Light Source Pro comes with, blocking out all extraneous light.

With the light source sorted, I'd already ordered a 4x5" LaserSoft Provia IT8 hand measured target as I mentioned previously. I established an ideal RAW exposure for this target in the rig, converted the ARW to DNG using Adobe's free tool, and then used that to create a custom profile in Lumariver Profile Designer Pro, on a freshly calibrated monitor.

I've now made test repros of sheets of Provia, Velvia and Ektachrome E100 with the new light and profile. Frankly I'm blown away. A *huge* improvement over what I was getting before. The Provia repro is 99% of the way to how it appears on a lightbox. The Velvia and Ektachrome repros are maybe a hair less faithful, as one would expect, but still 98% I'd say and absolutely good enough as a starting point.

@L Gebhardt you were correct to some degree about the light panel. The base result from the 99 CRI light is a notable improvement over that from the Kaiser, but the custom camera profile is ultimately where the biggest difference was made.

Being about 5 stops brighter than the Kaiser, the NS light also has the added benefit of allowing higher shutter speeds during capture. This appears to have essentially eliminated the edge artifacts that can result at 100% in pixel-shifted captures from the A7RIV due to micro vibrations at lower shutter speeds.

Overall I'm one damn happy camper right now, despite the lighter wallet

Adrian Bacon · Mar 17, 2023

GLS said:
Overall I'm one damn happy camper right now, despite the lighter wallet

Glad you got something that works for you. That's half the battle.

250swb · Mar 18, 2023

GLS said:
So an update on this.

I decided to splurge some more and do this thing properly. Bought the Negative Supply Scanning Light Source Pro 5x7" 99 CRI. Overpriced, but extremely well made and ideal for my existing setup; I discovered by happy chance that my custom-made 4x5" film scanning holder sits perfectly astride the built in plastic mask that the NS Light Source Pro comes with, blocking out all extraneous light.

With the light source sorted, I'd already ordered a 4x5" LaserSoft Provia IT8 hand measured target as I mentioned previously. I established an ideal RAW exposure for this target in the rig, converted the ARW to DNG using Adobe's free tool, and then used that to create a custom profile in Lumariver Profile Designer Pro, on a freshly calibrated monitor.

I've now made test repros of sheets of Provia, Velvia and Ektachrome E100 with the new light and profile. Frankly I'm blown away. A *huge* improvement over what I was getting before. The Provia repro is 99% of the way to how it appears on a lightbox. The Velvia and Ektachrome repros are maybe a hair less faithful, as one would expect, but still 98% I'd say and absolutely good enough as a starting point.

@L Gebhardt you were correct to some degree about the light panel. The base result from the 99 CRI light is a notable improvement over that from the Kaiser, but the custom camera profile is ultimately where the biggest difference was made.

Being about 5 stops brighter than the Kaiser, the NS light also has the added benefit of allowing higher shutter speeds during capture. This appears to have essentially eliminated the edge artifacts that can result at 100% in pixel-shifted captures from the A7RIV due to micro vibrations at lower shutter speeds.

Overall I'm one damn happy camper right now, despite the lighter wallet

Interesting how you got to where you ended up. But to be clear I assume you had the camera set to 5000K to match the Kaiser Plano and not on AWB?

GLS · Mar 18, 2023

250swb said:
Interesting how you got to where you ended up. But to be clear I assume you had the camera set to 5000K to match the Kaiser Plano and not on AWB?

Yes. White balance is/was fixed in-camera to match either light source.

GLS · Mar 18, 2023

By the way, I can confirm that the LEDs used in the NS light are the 5000K Absolute Series from Waveform Lighting. I had strongly suspected as much, and did consider buying some and cobbling together a box myself. However, the cost saving would not have been that significant and the task likely would not be a trivial one. Also I was just impatient to get things done.

250swb · Mar 19, 2023

It's often the fine tuning that makes the difference so good info, thanks.

Alan Edward Klein · Mar 19, 2023

GLS said:
By the way, I can confirm that the LEDs used in the NS light are the 5000K Absolute Series from Waveform Lighting. I had strongly suspected as much, and did consider buying some and cobbling together a box myself. However, the cost saving would not have been that significant and the task likely would not be a trivial one. Also I was just impatient to get things done.

I used my NEC puck used for calibrating my desktop monitor to measure my Kaiser Plano. Don't some cameras have that capability to register the "white" it's seeing and possibly give the reading in K Kelvins?

GLS · Mar 19, 2023

Alan Edward Klein said:
I used my NEC puck used for calibrating my desktop monitor to measure my Kaiser Plano. Don't some cameras have that capability to register the "white" it's seeing and possibly give the reading in K Kelvins?

Yes, the A7RIV can do this. It gave me a reading of 5300K for the Kaiser panel at full power. WB is only one part of the equation though; setting it in-camera (or later in post processing) cannot correct for discontinuities in the light's emission spectrum.

Alan Edward Klein · Mar 19, 2023

GLS said:
Yes, the A7RIV can do this. It gave me a reading of 5300K for the Kaiser panel at full power. WB is only one part of the equation though; setting it in-camera (or later in post processing) cannot correct for discontinuities in the light's emission spectrum.

What does that mean?

grat · Mar 19, 2023

They're referring to the fact that LED emission varies by wavelength. So does halogen. So does sunlight. Generally speaking for LED's the higher the CRI, the closer it resembles solar visible light, although there's frequently a bit of a dip in the "blue" range.

You'll only get actual discontinuity on CFL / fluorescent lights. Most LED's have full continuity at this point.

Alan Edward Klein · Mar 19, 2023

grat said:
They're referring to the fact that LED emission varies by wavelength. So does halogen. So does sunlight. Generally speaking for LED's the higher the CRI, the closer it resembles solar visible light, although there's frequently a bit of a dip in the "blue" range.

You'll only get actual discontinuity on CFL / fluorescent lights. Most LED's have full continuity at this point.

So what does that mean regarding using a Kaiser panel and the Sony A7RIV camera to read Kelvins? what does that mean vis a vis :scanning film?

Beverly Hills · Mar 19, 2023

Adrian Bacon said:
A simple way to do this is to take a photograph (on your slide film of choice) of an x-rite color checker card (using daylight, or flash), then digitize it with raw, then use that raw digitized frame to make a custom camera profile in Lightroom. There's no need to use an IT-8 target, those were for back in the day when people used scanners, not cameras. You can google how to make a profile with the color checker card. It's pretty straightforward.

Hello Adrian Bacon, would you agree with, that color correction like you described, is the basis correction from
raws ? If your way to digitize is then the same, you may
save this settings. From time to time you have to improve
your basic settings because from time to time there is a change in the workflow (the scanning workflow).
If you work with 2 scanning units you need two basic settings a.s.o.

2..) with lightroom or other products you come to the first flor of color correction (same procedure as in the darkroom)
but with digital filters.

but before this can be done you have to do

1.) your Monitor ( or TV sceen ) need its own calibration.
If you have color bars it would be fine in other cases you produce your own color test-chart.

If you have all this done correct you have neutral colors!!!

If that is fine for you then you have finished.

But with this saved data in true (uncompressed) raw format you may play your own (let us say ) game!

now you can change your colors how you feel fine with.
(that is the difference to darkroom color corection for me)

AND THIS SETTING CAN BE SAVED (Your personal Look/LUT) and with a little luck you can save a lot of time.
If your workflow is pretty correct your correction steps can be applied to full groups of shots.

Creating custom camera profiles for digital camera "scanning" of slides using an IT8 target: anyone tried this?

Moderator

Privacy & Transparency

Privacy & Transparency