BragIt
Image Fidelity

IT8.7 Profiling Tests

This page is devoted exclusively to problems when attempting to profile a professional-quality 35mm film scanner using an IT8.7 reference target and the scanner program 'SilverFast Ai 6' from LaserSoft Imaging. The quality of profiles generated by SilverFast Ai 6, inCamera 3.1 and Scarse 0.3 are evaluated, and examples of results for a portrait and a particularly problematic image are shown. The conclusion is that inCamera is capable of producing good profiles with full range slides, whereas SilverFast profiles destroy shadow details beyond rescue and creates new really nasty color casts!

The first experiments were published in 2003-12-29, and to some extent, that page is still relevant: the original article!. However, it is not so easily digested, and the most important information is reproduced in the new pages. Also, the IT8 module in silverFast may have changed somewhat since then. In addition, at that time I was not aware of the importance of the order in which you apply curves and convert to a working space, so some conclusions are more accurate in the current set of pages.

The whole issue of color management, profiles, white points and rendering intents is so complicated that I defer some of its intricacies to a page dedicated to: profiles, relative/absolute white points, rendering intents. The original page discusses several tests done at different gammas. The rest of this page assumes you have already acquired a fair amount of color management knowledge.

[Stockholm 2004-12-23]

For the latest findings, see an update to this article.

The setup

The film scanner used in these tests is a Microtek ArtixScan 4000tf, and the scanning software is SilverFast Ai 6 (version 6.2r3 was used in these tests) from LaserSoft Imaging (http://www.SilverFast.de) connected via FireWire to a computer running Mac OS X 10.3. Besides a target from Kodak (Q-60E3), which I will not use here, I have an IT8.7 target made by Wolf Faust from http://www.targets.coloraid.de. For studying the results, just about any imaging application can be used, such as Adobe Photoshop, Photoshop Elements, PaintShop Pro, Canvas, Gimp, ..., but for the sake of simplicity, I will sometimes use the product name Photoshop in place of any of these applications, or simply say "our image application".

As for monitor calibration, I use a 22" Mitsubishi Diamond Pro 2045u calibrated with PhotoCAL 2.7.7 from ColorVision, using gamma 2.2 (Windows users can therefore rest assured that this will give the same appearance as on a calibrated PC, despite this is a Macintosh). The 2.7.7 version of PhotoCAL corrects a very serious color cast bug in 2.7.6 (reported by me), and version 2.7.6 in turn corrects a minor white point problem with version 2.6. So now everything seems to be reasonably okay as far as monitor profiles is concerned.

Optimizing lamp lightness

It is very important to optimize the lamp intensity, which can be done from within SilverFast. The original technical specification for ArtixScan 4000tf (a pdf file) claimed to have a Optical Density: 4.3 dynamic range. First of all, that is contradictory and a damn lie, since optical density and dynamic range are not at all the same thing! In recent publications, they claim more correctly Dmax is 4.3. However, as it turns out, they mean that the scanner can only register light through a film with a density of 4.3 if the lamp intensity is cranked up to maximum! But if you crank up the lamp intensity, you completely destroy the highlights! So, in practice, you really have to work hard to capture the essentials of a full range slide, which may have a dynamic range of about 3.2–3.6 (which means up to 4000 to 1).

I have set the lamp intensity to precisely that level where GS0 (the clear film patch) of the IT8 target reaches RGB = 252-254. By this method, we have as high setting as we can get without clipping the highlights, while getting as good shadow performance as we can get. The exact lamp setting for my unit is: +8 units overall, plus an additional +6 and +4 units respectively for green and blue. I have tried to increase it even further (+10 units overall) so as to get RGB = 253-255 in GS0, but with the profiler inCamera, the results got worse in the super-highlight patches A13-A19. So lightness {8,0,6,4} is really the optimum for a Fuji slide in my scanner.

Warning: The User Guide for PictoColor's inCamera, version 3.1, has a severe error: they instruct the user to adjust lamp brightness so that patch GS1 will get RGB about 250. (That is detailed both in text an in an illustration). Don't do that! They actually mean patch GS0!

Optimizing gamma

The "neutral" gamma setting for this particular scanner is about 2.0. This means that at gamma 2.0, the scanner scans patch GS10 of an IT8 target to a "brightness" that is about the same as an IT8 calibrated scan would produce, using absolute colorimetry. In other words: at gamma 2.0, the image gets a brightness that is consistent with what an IT8 calibrated scan would produce (with absolute colorimetry.) This setting is sometimes regarded as the optimal setting (e.g advocated by Ian Lyons at Computer-Darkroom). This is the optimum setting if SilverFast is used for making a profile, but otherwise, if profiles are made with dedicated tools, it may not necessarily be the best. The best is to scan as raw as possible using HDR, (and in the first version of this article, I advocated using no gamma compensation at all, but see the update to this article for the latest findings, where gamma 2.0 is now actually the recommended setting.)

How the profiles were made

SilverFast Ai (http://www.SilverFast.de): SilverFast Ai 6 uses the preview of the scan to create a profile and set up a color managed work flow. In practice, it is based on only 8 bits per channel, and this is clearly insufficient to obtain good results. I can't see why a more accurate method wasn't chosen. [In July 2009 LaserSoft mailed me to say that SilverFast now uses 16 bits per channel. I have no information as to when the change was made.]

The SilverFast Ai profile was made at gamma 2.0 and lamp lightness settings as described above. All color management was first turned off, but the color space 'Adobe RGB (1998)' was thereafter selected. The 'limit gamma slope' check box was not selected. In Comparisons using the IT8 target you can see the result when the IT8 target itself is scanned using this profile in a color managed work flow of SilverFast.

inCamera (http://pictocolor.com): inCamera is a plug-in to Photoshop, and is based on full 16 bits per channel. It is intended for making profiles for both scanners and digital cameras. inCamera is distributed by PictoColor (the developer is actually Pictographics).

The inCamera profile was made by scanning the target at full resolution at gamma 1.0 and lamp lightness settings as described above, using the '48 Bit HDR Color' setting to get it as clean as possible. All color management was turned off, and there is no filter. In Photoshop, it is manually dedusted, and then the filter 'Dust & Scratches' is run with radius 1 and threshold 8 to remove the really microscopic particles that always stick to films. This will bring the image to a state that is as close as possible to the state one can expect it was when Wolf Faust measured it to generate the reference file. Then it was downsampled a factor of 2 to make it more "reasonable" in size. And as the last stage, a profile was generated with the Photoshop Plugin inCamera. In Comparisons using the IT8 target you can see the result when the profile is assigned to the IT8 target scan itself.

Scarse (http://www.scarse.org): Scarse is free open source color calibration and management software package under the hat of SourceForge (http://sourceforge.net) aimed at producing ICC profiles. It is still only at an alpha 0.3 stage, so it is not very easy to use — there is no GUI, and you have to run it from the command line! To install the package, you need to first download it plus a package called libtiff, and then compile it yourself, provided you have the right compiler etcetera. The package can read 48 bit TIFF images, and the internal precision is 56 bits per channel! The primary target group for Scarse is Linux users running Gimp.

The Scarse profile was kindly made by Andrei Frolov, the author of Scarce, based on a 1 Mpixel 48 bit TIFF scan I had done in December 2003 with SilverFast in "clean" mode, i.e with no color management, at gamma 1.0 and with lamp lightness settings as described above, using the '48 Bit HDR Color' setting to get it as clean as possible. In Comparisons using the IT8 target you can see the result when the profile is assigned to the IT8 target scan itself. In the original page I published on IT8 problems (it8testsOrig.shtml), you can read about tests I did with Scarse at other gamma settings. (However, I revoke some claims made there concerning "blotchiness", which I explain in a problematic image)

Comparisons of results for three images:

All links in the following list will open in separate windows:

For the latest findings, see an update to this article.

Harald E Brandt
Hägersten, Stockholm
Sweden
Photo: http://photo.bragit.com
BragIt: http://bragit.com
Email:
Last updated: 2011-10-27 at 15:07:10 +0200