Color Management/custom profiles at capture - slides

Just ran a test and D+ (highlight tone priority) does meter the image at X, then adjust to X -1EV only if needed.

I know I would much rather pull a shadow darker than push it brighter, especially considering color shifts. In my bracket tests so far, shadow noise when I equalized exposures (- 0 + EV) to match 0EV and then moved further with the shadow slider is visibly superior in the +1 (or more) exposure this way with no discernible differences in mids and highlights. But lab tests would say otherwise, I suspect. For critical single scans, I am using all manual mode in a completely separate setup/light source and a few test shots, etc., taking all of what I have learned here into deep consideration!

Again, quite the education few days for me.

For bulk scanning, my choice/theory involved a few things, all of which were based nearly totally on speed…

  • The noise at ISO 200/400 is functionally not all that different from 100 in practical real world terms.
  • The tonal/DR penalty is real, I understand. But… bulk scanning and then doing more critical stuff after makes this difference here less important.
  • It also gave me the shutter speed advantage I want because of the aforementioned projector fan vibrations!
  • We just can’t have it all, can we! I just want a perfect 100 CRI light, perfect flat field autofocus lens at 1:1 that can scan imperfectly flat film somehow at 2000+ frames per hour providing me a raw file at ISO 100, f5.6, 1/500th, 16bits and 100MP, with perfect color SOOC, hands free, every time without fail y’all! EASY!

It seems likely that you’re correct @nicnilov and I can just set min and max shutter limits to achieve the same purpose without the D+ limitations. More testing! This pushes me more towards to Solux bulbs since they are brighter and I can nullify the ISO 200/400 part of my equasion.

I cannot disable the projector’s fan and its vibrations, at least not yet that I can figure. I did remove the 82V/300W light and installed my own separately powered 12V/7.5W (single point LED) light so it can be on while the projector (fan) can be turned off for in the moment, vibration free scans. Though I lose the ability to advance the slides in this situation and thus bulk scanning becomes tedious, hence wanting to kill the fan entirely.

If I end up with the halogen bulb I will (perhaps!?) want to keep the fan for cooling, but maybe a Solux bulb isn’t so bad compared to the 150/300W bulbs it was designed for! If I stick with my current LED solution, I can truly investigate killing the fan but cannot harm the advancing mechanism. 12v/35W/50W Solux bulbs are going in for testing as soon as I get them.

In any case, the speed at which this can process, hands free, is hard to ignore.

For color and capture, CRI, R9, Lumens, and CBCP (center-beam candle power) measurements all matter. A brighter light is great… but I want to keep my lenses between f4-f8 for sharpness with limited diffraction and not have to use ND filters anywhere in the line.

Such a complicated and fun mess and I will keep saying it - Thank you all for such a lively conversation. I am learning so much about the speed, quality, and time cost compromises that are required!

AS FOR DXO DR and TONAL RANGE

I will do some testing for sure. And surprisingly DXO says R5 ISO 400 is better than 200 for DR but not for tonal range. So I tested it and 400 compares more favorably to 100 when equalized (+/-2EV for either) than 200 does (+/-1EV for either). So strange. I wish I had the resources ($) to test this all against a few other cameras.

I think this has something to do with Canon’s Dual Pixel technology (dual gain) which is hard to utilize efficiently (requires their software, DPP)…


Also, I am surely letting the perfect be the enemy of the good here.

Thanks for instigating this discussion about colour and colour profiling, very interesting. I’m thinking that you’re well on the way to creating a very good system with which to embark on your task. I’m slightly worried by your hot projector light source though. Solux Halogen bulbs are of course the gold standard but are they future proof? Clearly they’re still on sale but for how much longer I wonder. Nicnilov is absolutely correct to suggest that LEDs and CRI ratings are problematic but it would be great if there actually was an LED light source that would fit the bill. Certainly DT Transitions make one for demanding applications, as seemingly do Negative Supply so it must be possible. It would mean that you wouldn’t need to live with that heat, and that fan!

I’m not sure you really need to keep your shutter speed shorter than 1/125, with a short time delay and a solid setup I get sharpness indistinguishable from flash at 1/2 sec.

There is of course a good thread on light sources on here:

I have no knowledge of this company but they seem to be producing some interesting products.

I currently use a nice LED that runs cool enough to touch! The 35/50W Solux versions will be hotter but slides are never in for more than a minute or so max, but usually mere seconds. These both pale in comparison to the 300W bulb it typically works with!

As for shutter speeds, 1/125 is ok, but 1/250 or more is fractionally (but visibly) better so that is what I am going with currently.

I think I can adjust my light source to be more focused on the CBCP the bulb is able to do, but I need to get some scanning done first.

I’d be interested to see what Negative Supply is using honestly, because I can DIY-solder basic lights with the best of them! Ha!

Thanks for the links!

Well, I appreciate that you’re sharing all this information so I hope you don’t mind me suggesting that if there is a difference in sharpness between 1/125 and 1/250 then that might be an area to look at. On the face of it that would suggest to me that the camera support isn’t sufficiently stable and is allowing some kind of vibration to creep in.

Oh I understand! No worries at all.

It is the vibration of the slide projector turned scanner… its fan runs 24-7 regardless of need (or want currently) because its power is tied into the whole unit’s inner workings, including the most essential bit, the slide advance mechanism. It is all tied in to a Camerdactyl BoopBoop trigger which automatically advances the slide, fires the camera, and repeats. 1/125 is surely usable, I have even made prints up to 9x12" that look GREAT but since I can see the difference… knowing it is there means I am going to solve it since it is fairly simple, less unit vibration or faster shutter.

So I have wired/powered my lighting separate from the main power, and when needed I can turn the projector’s power off, thus killing the fan and vibration and I can go down to 1/60 if i want to (lower but the slides aren’t useful below that at this lumen/cbcp level). This is my main engineering issue I want to solve currently, aside from more technical color issues.

I am simply trying to optimize speed (projector hacked into scanner), quality (color/technical and sharpness/vibration), and cost (DIY, time, etc) to have a reasonable first look at this initial round of 10,000 slides or so, with another 35-40,000 waiting in line! After those, I move on to mass production of sleeved B&W/color neg contact sheets (but digital) before making final selects on those and then scanning them 1:1 as well!

It is quite the task, hence me REALLY REALLY trying to get this process of understanding and perfecting down pat. :slightly_smiling_face:

And as I go I want to be sure to share what I can if it is useful to anyone!

Thanks for giving your perspective on this, I’d always wondered what the difference might be, particularly with Kodachrome where targets are unavailable. Wolf Faust still offers that selection of 5 film bases, other companies seem to be simplifying their range although I see that Silverfast now offer a 3 frame ‘Advanced’ target, seemingly on Fuji or Agfa. There is a very detailed article on Luminous Landscape which uses it, I don’t pretend to have read it in any detail.

Thanks for explaining, that’s making a lot more sense to me now.

If we check what a human eye/brain combo can differentiate…any of the higher end FF cameras will be good enough.

If I remember correctly, we can differentiate about

  • < 240 hues
  • < 25 levels of saturation
  • < 500 levels of brightness

which corresponds to about 3 million colours or about 144 levels of r, g and b each - if sensitivities were equally distributed.

More bits prevent posterisation while customising and are therefore welcome. I’d not worry about something being suchandsuch number of bits, unless the values were really low.

Regarding projector vibration and exposure time, you might consider slide vibration due to mechanical shock (slide changer mechanisms aren’t that subtle) and decoupling ventilation from the light/slide/camera unit. How about sucking the heat out of the projector with a vacuum cleaner, attached with its flexible hose…and some additional inertia like putting the whole thing on a stack of garden tiles resting on something soft?

I must admit that I’d never heard of Ethan Moses, the BoopBoop trigger or indeed the Mongoose but he’s clearly a very resourceful guy. I watched the video on the trigger and he does show how to replace the bulb with a high CRI LED equivalent, even sells the bulbs and provides white perspex discs to replace the heat shield, but I’m sure you know that. Of course these bulbs are a bit of an unknown quantity unless you can find an absolute recommendation of one to do the job. I suppose the main problem is that even if you do that it seems you can’t disconnect the fan as you say.

You’re going to get a lot of balderdash here about profiling slide film. It would all be legitimate, if only you were working with fresh film and making your own exposures of test targets and profiling the resulting film.

My workflow is very simple. I make raw exposures of slides, and adjust them in Lightroom Classic. I’ll start with a representative sample of a particular emulsion, such as Kodachrome 64, do an exposure ring-around. From that, I’ll establish a basic raw exposure to use for “normal” slides. Basically, I then make adjustments to all the parametric sliders until my calibrated and profiled monitor image looks as close as I can get to the slide on a light table. I save that as a Lightroom preset, and apply it to subsequent slides. This gets me in the ballpark, with quality good enough to work with for most cull editing purposes. I’m using an EFH (Essential Film Holder) with a 2x2 slide holder.

I have to say, I am a former pro lab manager, used to scanning negatives by the millions, using $50K Kodak Bremson HR500 scanners. Even with that experience, I’ll say Negative Lab Pro is an impressive product for use with negatives. It does a great job of pulling things out of film I never would have printed optically. I don’t use it for slides.

For slides and NLP-processed negatives, I do apply profiles on OUTPUT from Lightroom. The entire workflow is in Lightroom Classic and Photoshop, using raw to PSD, raw to 16-Bit TIFF, or raw to DNG for intermediates (in ProPhoto RGB color space). Conversion to other profiles happens on output from Lightroom to printer, web, book, or exported file.

I’ve always found it a bit silly to apply a profile to old film, so I live with what I need to do in Lightroom with a raw file. Once I’m dialed in, it isn’t much!

With full control of the system - an excellent lens, calibrated light source, perfect alignment, optimum exposure, a solid vibration-free setup and time to clean and prepare each slide it’s pretty challenging to get optimum quality from, in my case, a 24MP sensor, much more so for 45MP. In addition it seems that even bracketing isn’t going to be possible for you let alone focus stacking.

However when you’ve perfected it your system should be able to do what you are requiring it to do (and I don’t know of another affordable system that could) but I think you might be setting your sights too high. I see that the R5 has a ‘crop-sensor’ mode for RAW files at a more manageable 17MP, that would surely be more than enough ‘to have a reasonable first look’ and your 70mm Sigma becomes effectively a ‘112mm’ with enough distance to reach the slide.

I think it’s inevitable that the best slides once reviewed and edited (and what a task that will be) will need more careful cleaning and re-scanning.

Going back to the O/P’s opening questions, there were two: (1) how to achieve high quality colour management for digitizing colour slides, and (2) whether the same techniques would be applicable to colour negatives. My post addresses those two questions, but there remains a lot to say about the digitizing set-up.

Starting with (2), the answer is “no”, as explained fully by Don Hutcheson here: http://hutchcolor.com/PDF/Cneg.pdf. I would also recommend Don’s paper on colour management basics here: http://hutchcolor.com/PDF/Principles_6.pdf.

Item (1) has many answers depending on how much money you want to spend getting it right. You can achieve reliable (for the same media batch) accurate profiles for your digitizing set-up and positive transparency film using high quality materials that also make the whole procedure very easy. It requires two main inputs: (i) a high quality profiling target, and (ii) high quality profiling software.

For the target, I highly recommend what a great many professionals use: Don Hutcheson’s targets - either batch-measured or custom-measured (less/more expensive). You can find them here: http://hutchcolor.com/hct.htm; they are called scanner targets but work very well for creating camera-based profiles. An X-Rite ColorChecker SG target may also be used: You can find that here: https://calibrite.com/us/product/colorchecker-digital-sg/.

For the profiling software, I recommend basICColor Input 6 - the non-pro version will suffice. There is nothing better on the market and it has well-designed workflows with instructions that are easy to implement. You can find it here: https://www.basiccolor.de/basiccolor-input-6-en-US/. People who digitize film for museums and other cultural heritage organizations use this combination of materials.

Alternatively, i would suggest looking at X-Rite i1Profiler for an integrated easy-to-use profiling solution, though it has become somewhat expensive and perhaps provides a lot more than you need.

Obviously very good software and targets of course and certainly some kind of reference using the same camera, lens & light source is required. However doesn’t the overall success of using profiles created in this way rather depend upon how you define your phrase ‘the same media batch’ in this context? He has upwards of 50,000 slides to copy, often presumably on old faded stock, and even perhaps unknown film stock.

Might not the advantages of using these relatively expensive targets and similarly expensive colour profiling software be lost compared with the alternative of a much cheaper IT8 from Wolf Faust or Silverfast and using the built-in plugin in either Lightroom or Photoshop?

Would a reflective target like the SG really be appropriate here? He would need to somehow light it evenly with two of the projector bulbs either side I suppose, and if he was using his LED light panel he would need two of these and use them in the same way. Clearly the film type is left out of the equation if profiles are created this way.

No. The quality of the profiling and the quality of the media are separate and separable factors. For the media side of it, the time-tested axiom “garbage in, garbage out” applies here in full force, except that with good colour management the garbage will have been properly colour managed - and this is not a joke, it is the essence of the distinction. You don’t want colour management to be the binding constraint on the quality of the final result, because one expects, especially with a large collection, that the quality of the media going in will vary a lot from being junk to very good, and you want/expect the very good stuff to come out very well.

On the targets, sorry if I caused confusion, but I should have mentioned that the SG is for camera profiling in the contet of photographing reflective media if one wants to do that.

The Hutchcolor targets are transparencies meant for creating input profiles of the kind useful to colour-manage the digitizing of transparent media. There are no targets available to match every film. LaserSoft Imaging for example needs to tweak their profiles for each scanner and film type for which they provide scanner profiles, which is one reason why SilverFast customizes their product per scanner and produces generally well-managed, predictable results. As one of the contributors to this discussion noted, he finds it useful/necessary to tweak results in Lightroom. I believe that will always be the case, but again, this is a reason to use initial profiling materials that will get you as far along the way as feasible with current materials and methods.

The very detailed discussion has provided some very good points. However, I’m not convinced creating a camera profile for a specific film target is helpful, at least as you get going first. Yes, the most demanding color matching for museums is as Marc_Segal says (and when he talks I listen very closely and learn.) I remember when digital cameras came on the scene for professionals, there was a lot of discussion on pro forums about camera profiling. For highly controlled museum/art reproduction it’s critical. But for photographers shooting people and products in studios with a number of different lights from the same manufacturer, the flash tubes of which were all aging differently (or the tungsten filaments of the lamps, pick your lighting poison), camera profiling did not yield better results. Often it was complicating and simply made things worse, depending on the software. A lot of discussion surrounded the approach by Adobe, with a generic profile of your specific camera model created from two measurements at both daylight and tungsten. The Adobe method was hotly contested by some of the experts at the time, but ultimately Adobe won out for the vast bulk of professional users. If you are using Adobe Camera Raw/Lightroom, or Capture One, you already have a very good profile of your camera built in; not perfect, but save your special super-accuracy demands for after getting the project going with all your lighting, lens choice, film holding, mechanical and alignment needs worked out. For your project with different film batches, aging, different emulsions, variations in set up, variations in film processing, an IT8 profile is going to be limiting. The IT8 target (Ektachrome) does not have the full color range of Kodachrome or Fuji films, let alone the range of your camera sensor, but I believe applying the profile may limit the color range (out of gamut color will be mapped/changed to the profile color space limits). Even the superb Hutchinson targets may limit the color range of the digital sensor, though I am not qualified to test or determine that. My recommendations are (1) Use the highest quality lighting you can find; (2) White balance to your light source and make a flat field/evenness ‘profile’ in Lightroom/Capture One/your specific software at the start of each scanning session; and (3) If possible use a linear curve and set exposure using the linear curve. As of the last time I used Adobe Camera Raw, a gamma and LUT is always applied, so you can’t use linear capture; the linear curve they offer is AFTER the camera profile is applied with its gamma correction and LUT applied. I use Capture One. You can tame extreme contrast in slides by using a linear capture, and only modest adjustments to the raw file will yield really great results. As for high CRI lighting, the most important consideration is the R15-patch CRI. Here is a discussion from Waveform lighting: Ultra High CRI LED Lighting | Waveform Lighting I use the Waveform LED products but have no affiliation with the company. And remember you are diffraction limited past f/8, and at 1:1 reproduction your lens marked at f/4 is actually at f/8!

I linked to what I described as ‘a very detailed article on Luminous Landscape’ without actually noticing that you actually wrote it!. My apologies, it really is a very good article.

The context of this discussion is not that. The context is digitizing film. The variables are fewer than what you mention and controllable, so that profiling can work.

Without evidence this statement is really open to question. The dynamic range of the sensor is one factor, the other being that of the media being digitized. The binding constraint on the usefulness of a profile’s coverage is the lesser of the two. To analyze this proposition you would have to know how many colours and which ones in the Hutchcolor target are at or beyond the gamut of the sensor or the media, and you would need to know enough about the profile calculation methodology to understand whether the interpolation and projection of all those colours going into creating the profile yields a large enough gamut to at least match what the sensor or media can yield - because in this case the 3D-gamut map is more important to measure than the DR. But there is a way of finding out - you can analyze the profiles themselves in ColorThink Pro, analyze the colours in the media in ColorThink Pro, then map the latter into the former and see whether there are colours in the media, including the grayscale, that over-spill the gamut boundary of the profile. If you have relevant evidence that would be interesting to see. Meanwhile, anecdotally, I have seen professional results using these profiles with high-end digitizing equipment and not observed any remarkable colour clipping - in fact results out of the box for digitizing transparency film were very good.

I’ll close just by noting once again that problematic media is not a reason to avoid the use of high quality profiles for digitizing positive transparent media.

1 Like

Thank you Harry, I appreciate that. It was an absorbing bit of research to delve into! :slight_smile:

The initial context of this discussion is color management in digitizing tens of thousands of slides in a workflow designed for speed and good but not perfect reproduction. The methodology suggested to determine whether a profile will clip color from either the camera gamut or the media gamut is no doubt spot on and I acknowledge Mark Segal’s mastery of this topic. I thought it was clear I was speculating about any limitations from using the highest level of profiling film targets to profile a digital camera, not asserting there indeed is any. Tip of the hat to all.