What happens to a scan of a color negative when you change the color temperature and/or the color reproduction of the scanner’s light source, and how can you get the original colors back again?

This is a question I am facing now after replacing the fluorescent tubes in my Imacon Flextight Precision II negative scanner this morning.

Until recently I have used the original light tubes that came with the scanner when I bought it secondhand a few years ago. These tubes still function but I decided to change them as it seemed that the scans I was obtaining from my 4x5 negs reproduced less detail in the shadow areas than I had hoped.

To change the tubes is no big deal as the procedure is outlined in the user’s manual. But of course the exact model of the original light tubes specified in the manual (Osram L8W/12-950 Lumilux de luxe Daylight) are no longer available. So I bought a pair of new model tubes (Osram L8W/954 Lumilux de luxe Daylight) that as you can see have very similar specifications. Much to my delight the new tubes fitted in the scanner and produced an almost identical scan to the old tubes. But the color reproduction was slightly different.

Concerned by the color change (I have about 400 negatives waiting to be scanned), I wrote to Osram support to ask if there was any difference between the two tubes. This morning I received this thought-provoking reply (which I translate from Danish to English):

Hello Peter
The color temperature and the color reproduction is not the same.
L 8W/12-950 has color temperature 5000K and color reproduction (farvegengivelse) is approx. 98
L 8W/954 has color temperature 5400K and color reproduction (farvegengivelse) Ra > 90, approx. 92
This means that you will probably need to undertake a big calibration job.
Regards,

Consequently, I am faced with the following questions that I am unable to answer so would appreciate any insight or suggestions you may be able to offer.


1) What difference can I expect to see in a scan of a negative if the color temperature (in this case of the scanner’s light source) is increased from 5000K to 5400K?

2) Does such a change in the color temperature of the scan impose a limit on the output quality of the color reproduction or does it just mean that I have to work a little more on adjustment of each image ( eg in Photoshop) to make it look right (whatever that means!)

3) What if any significance does a change in color reproduction (farveangivelse) from approx 92 to approx 98 have?

4) Could these differences in color temperature and color reproduction be corrected by running the scans through a raw conversion software like Capture One that enables adjustment of the color temperature. If so how do I convert my TIFF scans into RAW format (Imacon Flextight uses its own proprietry raw format)

5) What could be meant by a “big calibration job”? Does this mean I should try to recalibrate the scanner (sounds challenging) or find a way to use software such as Photoshop or Capture One to “recalibrate” the scans?

6) If you were me and had time to fool around with calibration - but not money to buy a new scanner. What would you do?

Hope this is not too many questions!

In case of CCFL correlated color temperature doesn't tell the whole story - even when they're the same, their spectras may differ a lot. As a result, you may get different color rendering of certain colors. The only solution I know is to use IT8 or HCT targets and profile the scaner - but the problem is, that negative scaner targets were not very common, so there may be a problem to get one...

There are no specific targets for profiling a scanner to handle colour negatives. This is a well-known problem - there are simply too many variables with these materials. You didn't mention what software you are using with the scanner. The best solution I've found over the years doing this is Silverfast Ai Studio 6.x available from Lasersoft Imaging. It contains a utility called "Negafix" which allows you to create presets tuned to the kind of negative material you are scanning. You need to use a reference image with good memory colours (I recommend Bill Atkinsons PRinter Test Image available for download free on his website) so that you can see which Negaifx settings give you the most satisfactory reproduction of your negatives. Once you have this preset nailed, you just use it for all images shot with the same material. I've written a couple articles published on Luminous-Landscape about scanning negatives using Silverfast.

I believe that you probably can accomplish the effect of Silverfast "Negafix" using your existing film scanner software and Photoshop. Silverfast "Negafix"adjustments are made to the scan data after the scan is made. Instead of using Silverfast to make those adjustments, you could create and save a curves adjustment or most any other image adjustment you desire in Photoshop, and then apply those adjustments to your scans. In addition, such adjustments can be on layers, so you can tweak the adjustments for each film frame if necessary without any additional image derogation.

Actually, at least Fuji had IT8.7/2 negative version target supported by Fujifilm ColourKit. You could profile your scanner with that target, and then the software was somehow adopting to various negative emulsions.

No. Negafix is a set of canned or custom-made profiles selected before you scan the image, and when you scan, the scanning calculates tone and hue according to the selected profile; if it's correct this will maximise dynamic range and satisfactory colour appearance. While you can simulate almost anything you want in Photoshop, the point of using Negafix is to start with an approximately correct, optimized image from the get-go at the scan stage, most likely reducing the extent of further editing in Photoshop, which depending on how it's done, could be not as "non-destructive".

While I'm not familiar with this software, I'd be leary of the "somehow", because each negative emulsion needs its own "profile".

Mark, I guess we’ll just have to agree to disagree. You seem to imply that Negafix somehow affects the scan itself. It’s physically impossible for Negafix to operate in that manner. If I misunderstood you, I apologize.

Nothing to apologize for - we simply have a different idea about what's going on "under-the-hood". Could you explain why in your view it's physically impossible for Negafix to operate in the way I suggested?

Thanks for the input.

After reading your discussion about targets and calibration tools that can be applied "under the hood" of the scanner (ie to be applied automatically to every scan?), I am jumping to the conclusion that using light tubes with a slightly out (ie +400K) color temperature will not "harm" the quality or quantity of raw data captured by the scan. It will just mean that what I get on the screen when I first open up the file will need some tweaking (RGB or CMYK curves, levels etc) to "look right". Is this conclusion correct?

If this conclusion is correct then I can breathe a (small) sigh of relief. At least it indicates I can continue to work with this particular scanner and the new tubes instead of having to junk the whole system. It also implies that if I do continue to work with this scanner and the new tubes I will have to take the time to correct the color of every negative that I scan.

As you mention it would be very good if the color correction can be "automated" by using a predetermined target profile that I could automatically apply to every scan instead of having to reinvent the (color) wheel every time. If I cannot find a program like Silverfast to do this for me then maybe I could develop my own "ballpark target" by adjusting levels and curves then saving it in Photoshop and applying it to every scan.

My normal workflow is to spend up to one day working on each scan to get the color of the image where I want it to be, so am used to working with curves, levels etc. To scan I use Imacons Flexcolor software. Usually I just scan using Imacons proprietry raw format (3F .fff) to capture as much data as possible then export from Flexcolor to Photoshop in 16-bit TIFF format.

So my only real concern is what do I stand to lose in terms of data harvest by scanning with the new tubes instead of the original ones - apart from time required to adjust curves, levels etc.?

Maybe nothing at all?

The flexcolor program does give me the setup option to micromanage the specific color array of the scan (increasing or decreasing color values) but I am not sure (a) how should CMYK/RGB colors be adjusted to compensate for change in color temperature (sounds very complicated) and ( if this would make any difference to a (3F) raw scan anyway?

Any ideas about roughly what colors are most likely to be influenced by a minor increase in color temperature?

Which brings me to my last question. "In case of CCFL correlated color temperature doesn't tell the whole story - even when they're the same, their spectras may differ a lot. As a result, you may get different color rendering of certain colors." What does CCFL mean?.... It seems that "different color rendering of certain colors" is precisely the problem I am struggling with.

CCFL stands for Cold Cathode Fluorescent Lamp - the light source of your scanner. The color temperature is only a precise description for radiators - black bodies heated to a certain temperature. The specta of CCFL is different - for example I measured 6500K Philips Graphica Pro 950 CCFL spectra (red chart), and compared it with 6500K D65 illuminant spectra (black chart):

As you can see both lightsources have the same color temperature, but different spectras - so they will render colors in a slightly different way. It's hard to predict what colors will be affected in case of your light source, and the distorsion probably may be too complicated to correct it with some simple Photoshop curves. The simpliest solution is to buy a scanning target

that contains many color patches of known colorimetric values, and create profiles, that will contain correction tables. The only problem is, that negatives were not stable and too tricky to profile, so negative targets were not very common - it's only easy to buy targets on transparencies and reflectives.

Correct, and Silverfast's Negafix is the most practical solution to this problem - if there is a version of the program adapted to your scanner.

Silverfast does not support Imacon.

The flexcolor software does some kind of generic color correction to compensate for the emulsion color when it is inverting a negative. And the negative "film profiles" are just settings for the levels correction. These you could also do in photoshop with a levels or curves adjustment.

For calibrating the scanner, you need a scanning target on positive film, and a calibration software to make a profile for the scanner.

After reading the above comments it became clear to me that everybody was talking about "targets" so I thought that at least I should try to find out what one was and how it is used. While hunting I found the following quote about scanning color negatives from a book "Real World Color Management" by Fraser, Bunting and Murphy on another forum, which suggests there are several obstacles, namely (to quote)

* Nobody makes a color-negative scanning target.

* Unless you like orange, inverted images, you don't want to reproduce what's
on the film.

* The orange mask on negatives varies so much with exposure that, even if
someone did make negative targets, they'd only work if your negatives were
exposed the same way the target was.
</end quote>

I've seen flows that make a valiant effort by printing from the negative scan
and trying to profile THAT, if you want to try it. Bon chance.

After reflecting on this for a while I looked up my Imacon Flextight Precision II scanner manual and found instructions for white balance calibration of reflective images with the recommendation that this should be done every time the tubes are changed. So I did it (with a white piece of A4 paper and no plastic sleeve ) though of course! this seems to have made no difference to the negative scans (which are done by a different tube than the one used for reflective scans). Perhaps I also managed to completely screw up the calibration of my scanner for reflective scans but that is another story and no big deal since I never do reflective scans anyway.

Yesterday I found a Photoshop plug in by ADG that adds to filters in Photoshop a sexy plug-in with a color temperature slider to adjust the color temperature of TIFF and other files in Photoshop. When I used this plug-in to reduce the color temperature of the image from negative that I had scanned with the new tubes (5400K) by 400K the colors did actually look more similar to the colors of the image from the same negative scanned with the original tubes (5000K). But strangely enough both images appeared to be much too yellow (even though I had carefully calibrated my monitor with a spyder) to begin with, indicating perhaps that the color temperature was too low on both scans (could this be caused by old tubes).

Consequently, the most "natural" results that I have been able to get so far from either negative scan (old 500K tubes or new 5400K tubes) were not by adjusting the color temperature using ADG's Photoshop plugin but by using Photoshop CS3 IMAGE > ADJUST > photo filters > 80 or LB cooling filter. I guess this is just a different way to simulate reduction of color temperature via adjustment of curves and levels than the ADG color temperature plug-in rather than actual filters - but at least they provide a standardised profile that I can apply to every image I scan as a starting point for adjusting the colors.

As a result I thought that I had found a way out of the jungle - until of course I looked carefully at the very interesting chart Czyorni posted and read these conclusions, which seem alarmingly logical even to me:

"It's hard to predict what colors will be affected in case of your light source, and the distorsion probably may be too complicated to correct it with some simple Photoshop curves. The simpliest solution is to buy a scanning target that contains many color patches of known colorimetric values, and create profiles, that will contain correction tables. The only problem is, that negatives were not stable and too tricky to profile, so negative targets were not very common - it's only easy to buy targets on transparencies and reflectives."

One steps forwards, two steps backwards...

Well at least I now know more than I did before but perhaps it is true that a little knowledge is dangerous.

It should be normally less destructive to correct as much at the scannng stage as possible. Calibrating a device means putting it into a known and repeatable state by setting it's basic parameters in the scanner driver (black point, white point and TRC) - not scanning targets. Once the scanner is calibrated, the targets are used for profiling, and positive targets will not generate correct profiles for negatives.

Mark, for purposes of our discussion, a scanner has two essential parts, a light source and a light sensor; e.g. a cold cathode fluorescent lamp (CCFL) light source and a CCD light sensor. For Silverfast or any other scanning software to affect a scan as you suggest, the scanning software would have to be able to control and adjust (1) the spectral output of the CCFL and/or (2) the way the CCD responds when exposed to light. Scanning software can do neither.

Mark, once again, we have a difference of opinion. Except for hardware adjustments (e.g. exposure), all scanner software corrections are made to the scan data after the scan. Those corrections are as destructive as similar corrections made in Photoshop. Moreover, scanner software corrections can not be done on layers. There may be reasons why someone would prefer to use scanner software to make corrections, but I prefer to make my corrections in Photoshop on layers for the same reasons that I’d rather use layers in Photoshop instead of directly adjusting the background layer.

Of course, if someone has an old scanner that records data in high bit, but outputs only 8 bit data, then there may be some benefit to making corrections with scanner software if the corrections are made using the high bit data. In addition, my comments relate to typical home scanners, not necessarily drum scanners.

There are conditions you can correct for at the scan stage which are much more difficult, if not impossible, to similarly correct in Photoshop - very severe exposure corrections being a case in point - from personal experience. I shall have verified the question of processing sequence and relative destructiveness.

Thanks, you may be right, but I think there is more to it; I'll obtain further input on that and report back when possible.

Some of the other CCD/ tube CCFL light scanners have a calibration kit , such as Creo Scitex. Changing the bulb is sure to have different characteristics for which there must be a way to balance the whites. On drum scanners we balance the drum and or (rather with or without film and mount) at all the apertures for consistency of the light. We usually do this every session as the temp of the PMTs are sensitive to environmental conditions.


I don't know that scanning clr negs though would benefit that much as they are really unique clr docs each having a different mask and density depending on the scene and processing and type of film.

IF you had old presets that were reliable, then a light source calibration would bring it back in line with before changing the bulb, but that is apparently something that is not a user function on Imacon.

Hope you are well into some successful scanning by now.

I also use a Flextight II and have a couple of thoughts on your dilemma having just had the tubes out of mine.

I bought the scanner new a good few years ago and have never had need to change the tubes. Two years after it's purchase I took the digital route for my commercial work and have only used it for scanning personal projects since. I have never scanned reflective originals (as it's such a pain with this model) so the top tube has never been used. They are the same model, they degrade with use as opposed to age, so earlier today a simply swapped them over (after a wipe down with some Isopropanol). This should extend the scanners life for a good few years with no expense or stress.

With regard to your colour change, I usually scan transparency where I take a grey (ish) reading and then adjust to taste. On the few occasions I've had need to scan colour negs I've found the canned 'profiles' able to only get me on the playing field, I always make adjustments in the 'Colour correction panel' and then save them for further use via "File, Setup, New". They then appear in your Setup list to repeat on the same neg type - no need to correct every image individually.

I believe that the scanner does take a 'white balance' before every scan (it is done through that cut-out you see above every frame aperture in your film holders). So the colour shift you're experiencing must be a mismatch between the spiky spectrum of light from your new tubes and the profile Imacon made for your neg type. Again, do colour correction and save a preset.

Though I've read comments to the contrary on this thread, it is my understanding that as much correction as possible should be done in Flexcolor at the scanning end as opposed to Photoshop. This is a high end scanner which cost £9000 back in 2002 (the latest version about the same now), the software is very sophisticated and designed for reprographics and commercial photography. Making changers to the preview image is a little like adjusting the Raw file before outputting a Tif. You can in fact scan to "3F" which is raw data, then do your adjustments later (but the files are huge).

Hope you've sorted things out in the meantime.