Scanner profiling options

[Blasted]

Hi there,

A while back I got myself a reasonably decent monitor (Dell U2713H) and to make full use of the monitors calibration capabilities I also bought an X-Rite i1 Display Pro, Dell made sure that's the only one that will work fully with this monitor and uses Dells Ultrasharp Calibration Solution, a version of the x-rite software specifically for the Dell's.

So now I have a film scanner (screen cezanne with powermac G4) and would like to profile that as well.

The Dell software allows display and projector profiling. It also has printer and scanner options but these are locked as demo's. As I understand it, I need to have the correct licence to make use of the scanner profiling feature and I also believe this is a mega expensive upgrade. The only other way would to be get iPublish (very expensive and more than I need or the i1 studio, which i'm not sure if it will work with this monitor).

Are there any other software packages out there a beginner could use to say scan a target and create an ICC profile for editing film scans in lightroom/photoshop?

Thanks.

[Peter De Smidt]

Hi Ben,

I'm not an expert on this, but I do have a Cezanne, and I have profiled it. (I also have an NEC monitor with i1 Display pro.) There are a lot of options, but first, you're scanning slides, i.e. transparencies, right? Profiling doesn't really work for negative film. Assuming that's right, you'll want a color target on the type of film that you want to scan. See: http://www.targets.coloraid.de/ You can then use L-prof or similar to make an icc profile. See Wolf's site. Perhaps email him to see what he currently recommends. You then make a scan with everything software adjustment turned off.....I can tell you about that on a Cezanne when you get to that point. Take it into L-prof. Create an Icc. Load the icc into the proper area on your system. To test, open the file into Photoshop, you assign the icc profile and then convert to your editing space. When you assign the profile, the colors should look better. Don't ever save this profiling scan again, especially with an embedded profile. Just leave it alone. Now scan some of your film with the same settings. Open a scan in Photoshop. Assign the icc profile. Convert to your working space, the later should be Adobe 1998, Ektaspace, or Prophoto RGB. The colors should be a faithful rending of the slide. Adjust to taste.

[jim10219]

I use two programs that work together called DisplayCal and Argyll CMS GUI with my i1. It allows for monitor, printer, and scanner calibrations, and they're free. The only downside is they're not easy to use. DisplayCAL isn't too bad. It works like a regular windows program. But Argyll CMS requires a lot of DOS command line type of stuff. There are some websites out there with step by step instructions, so if you do a search for them, you can get by without any serious computer knowledge, but it will take a lot of time and patience. That's the price you pay for free.

Though to be honest, with film (negative or transparencies) I don't think scanner calibration is important. You'll want to adjust the colors manually for each photo (or I do anyway) so scanner calibration doesn't do me anything. The monitor and printer calibrations are pretty important though. If I scanned printed photos more often, I would probably go ahead and do the scanner calibration. It would save me the step of having to match my screen to actual print, or at least shorten that step.

[Bernard_L]

There are a lot of options, but first, you're scanning slides, i.e. transparencies, right? Profiling doesn't really work for negative film.
Depends if you want to see on-screen the colors of the scene or of the film. Granted, colors of the film makes no sense for negative. As far as I'm concerned, my goal si scene colors, and I profile jointly my film and my scanner, using
- Coloraid reflective target, shot on film (preferably on each film)
- Argyll to generate a LUT profile (with extra headroom on the white end)
the downside is that such calibration is not transferable to someone else, but does anybody care? Another downside is that this workflow (aimed at scene accuracy) makes, at least in principle, all films look alike; so forget that Velvia look or that Ektar look, etc...

I have notes for my workflow (including those horrible DOS-like commands), which I'll be happy to share after 15-Sep (away from home just now).


I don't think scanner calibration is important. You'll want to adjust the colors manually for each photo
That may be fine for correcting the balance between the three primaries. But what about unwanted couplings between primaries; requires a 3x3 matrix, can you play that by ear?

If you want to preserve the "look" of each film, yet achieve color balance without the guesswork and time spent in manual adjustments, just shoot on film, instead of a coloraid target, a gray card, and use that to batch-correct all similar (same film, lighting, processing, scanner) images.

[Ted Baker]

Profiling doesn't really work for negative film.

That's certainly true if your talking about the it8 targets and your using them for what they were designed for. Of course that doesn't stop people adapting those techniques for better or worse, but I might suggest that's not a beginners activity nor are you just profiling the scanner and film combination any more... Not that should never stop anyone... :o

[Blasted]

Thanks guys,

I typically only scan colour negatives. I appreciate profiling is better for slides and such but assumed a basic profile would have been better than nothing? I would probably get better colours if I were to understand lighting conditions better (I try my best) and use exposure compensation more. for example i recently shot some portra in quite bright sun and for the most part the images were bluer/colder.

In short I've always been rubbish with colours and anything to give me a helping hand is most welcome.

[Peter De Smidt]

I don't think profiling is the way to go in that case. I'd work on making your workflow better. You might post an image as a sample. What is your process of scanning with the Cezanne?

[Blasted]

Please excuse my ignorance, but why would profiling not be good for colour negatives?

I will have a look and see if i can pick out an example.

[Bruce Watson]

Please excuse my ignorance, but why would profiling not be good for colour negatives?

Due to the much larger dynamic range of color negatives, and the nature of the color correction mask.

Trannies OTOH, have a much more limited dynamic range (or density range if you'd rather) and are WYSIWYG, which makes it much easier to plot a reasonably accurate profile. Which might, or might not, be good enough for whatever someone is trying to do.

[Ted Baker]

Please excuse my ignorance, but why would profiling not be good for colour negatives?

I will have a look and see if i can pick out an example.

In the case of it8 targets is just does not make sense in terms of what they are designed to do. If you take a step back it may help to understand how they work and what problems they are trying to solve. In terms of colour if the scanner sensor matched the response of the human eye, or more correctly the "Standard Observer" and the light source was identical to some standard, then there would be no need to calibrate the scanner. A certain green for example which is created with a mix of two or more dyes illuminated with the correct light could look identical to some green object seen by our eyes yet a sensor may see the green created by the CMY dyes differently than our eyes. This difference can be made worse if the illuminant is not broad spectrum.

The scanner designer should understand this stuff, and the calibration helps to reduce these errors. That is why you need to calibrate for each set of dyes. The target has a range of colour created with the correct set of dyes, e.g. ektachrome. Of course the actual design of the software should take some of this into account in the first place.

In the case of negatives it is much more complex, first the spectral characteristic of the paper do not intentionally match those of the human eye. There is no need to match the spectral response and this helps to solve some of the problems created by using imperfect dyes. The red and green layers are slightly more sensitive to shorter wavelengths for example.

In addition what is recorded in a negative is not just the colours "upside down" in an image. They are stored if you like with low contrast, ready to be expanded by the paper. It is this system working together that actually needs to profiled. How this is done, is up to the software solution you use.

I am certain there are other profiling techniques for colour negative, Kodak and Fuji would have a mountain of info on products they designed and produced, some of that logic is embedded in many products. The Fuji Frontier is one example.

[Peter De Smidt]

Good summary, Ted.

This is just an idea. I haven't tested it. For color negatives, you could take a picture of a MacBeth color chart in representative light. One problem is that you'll have to choose to treat it as illuminated by the main light or the fill light. So, if you go out on a clear day and illuminate the chart with the sun, the darkest square will be much brighter than the darkest area in a real scene, where the darkest area will also be in shadow. Personally, if I shot a lot of color negative, I'd shoot a MacBeth in both sun and shade, using the same exposure for both, ideally in one exposure, which would require two charts. Another option would be to make a shade for just the darkest tones of the chart. Scan. It doesn't have to be super high resolution. Take the scan into Photoshop. Compare readings to those on Babelcolor. Can you make adjustments in your scanning software, preferably hardware adjustments, that lead to a better scan? If so, do so. Now make three adjustment layer curves in Photoshop. Label the top one "color" and change it's blending mode to color. Label the middle one "luminosity" and change it's blending mode to, you guessed it, luminosity. Label the lowest one "invert". Drag the ends of curve to the other corners of the graph. This will invert the image. Go into the properties of the luminosity layer. Use the white eye dropper on the white square of the chart. Use the black dropper on the shadowed darkest square. Now try the mid point on some of the middle gray squares. Pick whichever one looks best, assuming you have a calibrated and profiled monitor. Now go into the properties of the color adjustment layer. Do the same thing. Your color should be pretty good. Look at the other colors and see if any need adjustment. Use any of the adjustment layers, such as spot color, hue/sat...., to get where you want. Now save all of these adjustment layers as presets. Or open this file and drag the layers to future scanned negatives. These will be just starting points.

Note, this would be an attempt to get accurate color. That would minimize difference between films. Maybe that's not what you want. In which case, I's simply neutralize the grayscale, not bothering with spot color/hue/sat.....

[jim10219]

Whenever I scan a photo on a new type of film and get a result I like, I save the curve preset. Then, when I rescan that type of film, I reuse that preset, at least as a starting point. It might go through a couple of revisions before I settle on a more permanent curve preset.

I also made presets for each type of film in my scanning software. I use Vuescan. They’re pretty basic though. They’re basically things like multipass for slide films and color adjustments to keep all of the colors in the middle of the histogram. I’m not trying to do any major edits here, just get something I can work with. I’m just trying to create a base file that my PS presets will work with.

With those basic settings, I can usually have a pair of photos color corrected in Photoshop before the next pair of photos are done scanning.