BragIt
Image Fidelity

IT8.7 Profiling Tests — Continued

This is a continuation of the "IT8.7 Profiling Tests" article — click here to go to the start of the article, which describes details of the setup and how the profiles were made.
You may also need to read about profiles and relative/absolute rendering intents of white points.

[Stockholm, December 2004]

Comparisons using the IT8.7 target

The IT8.7 target

The following images are of an IT8.7 Fuji Astia target made by Wolf Faust from http://www.targets.coloraid.de. All images are scanned in SilverFast Ai 6.2.3r6 at full resolution and then downsampled. There is no sharpening whatsoever involved in the chain. For information on how the profiles were made, see the start of the article.

Settings

The setting in SilverFast called '48 Bit Color' is used when you want SilverFast to handle color management, and, if you like, other manipulations. The setting in SilverFast called '48 Bit HDR Color' is the setting used when you want as "raw" unmanipulated results as possible, usually at gamma 1.0 and usually with no embedded color profile.

Web adaptation

Just prior to saving these images in a web friendly format, all of them have been converted to the sRGB profile, using the relative colorimetric rendering intent, and the profile is embedded in the images. The conversion engine is Adobe (ACE), but tests confirm that the difference between this engine and e.g Apple CMM is very small. Then they are converted to 8 bits/channel and jpg-compressed with quality setting 'High' (8) in Photoshop. Downsampling to small thumbnail versions is done in 16 bit mode.

The fact that they are in 8 bits/channel sRGB and jpg-compressed might have you suspicious, but when a scanned and profiled IT8 Fuji film target is converted to sRGB, the only colors that change significantly, as seen on a calibrated CRT monitor, are the area F3–I4, the area F7–H8, G12, and the pale patches in area A13–A19.

Your browser may or may not understand the sRGB profile that is embedded in the images, so to be on the really safe side, you need to download the images to disk and view them in your icc-aware image application. Tests on a Mac OS X system shows, however, that the appearance of these images in Safari (1.2.4) is precisely the same as the appearance in Adobe Photoshop!


SilverFast Ai 6.2
it8-image
Click to see a larger version in a separate browser window (120 kB).
This image is scanned in SilverFast Ai at gamma 2.0 with the setting '48 Bit Color', and it is color managed by SilverFast using a profile created by SilverFast, which converts the image to the standard working space Adobe RGB (1998).

(And as all the other images here, it is converted to 8 bits/channel sRGB and jpg'ed at quality setting 8.)

Overall, it does not look so bad, but if you look close, you can see that the grey patches are not neutral, and the darkest patches are wrong in all three Lab coordinates.

For some strange reason, despite the fact that the rendering intent is set to relative colorimetric in SilverFast, it does not map the white point to the one intended for 'Adobe RGB (1998)'; the result is that the highlights of the image look dark. As explained in Absolute/Relative Intent, White Point, a relative colorimetric intent should make patch GS0 close to 250-255 in RGB values, but the SilverFast version makes GS0 to about [222, 223, 227] (as seen in sRGB). The mapping seems to be absolute in terms of Lab values, but relative in terms of color temperature. However, this makes it possible to directly compare the Lab values with those in the reference file:

GS0 for instance (the clear film patch), gets Lab coordinates [89, 1, 1], which is about 1 or 2 units off from the reference file value [90.49, 1.03, 1.05]. Not much to worry about, but:

GS22 (the black just to the left of the white line) should be [1.64, 0.66, -0.05] in Lab coordinates. Measurements show about [5, 2, 2], however, so not only is it too bright, it is not neutral. The difference in 'L' between patch GS22 and GS21 should be 2.0, and another 2 units to patch GS20. With the SilverFast profile, however, we only get a difference of about 1 unit between these patch pairs, so the shadows are heavily compressed (and not neutral!). This is also precisely my experience when scanning slides! (All these values were read before sRGB and 8 bit conversion, since it sometimes differ about 1 unit and I want the most accurate ones.)


inCamera 3.1, relative colorimetric
it8-image
Click to see a larger version in a separate browser window (140 kB).
This image is scanned in SilverFast at gamma 1.0 with the setting '48 Bit HDR Color', dedusted, assigned a profile produced by inCamera 3.1.

It's a really good result with no artifacts! If you compare this with the SilverFast version, you can see that the very pale patches A13-A19 are much better here — bright but still with some color! In the SilverFast version, the patches are almost washed-out (despite they are very far away from RGB 255) — they should have been both brighter and have more color. inCamera makes them just about perfect!

inCamera 3.1, absolute colorimetric
it8-image
Click to see a larger version in a separate browser window (132 kB).
This image is converted so as to maintain the Lab values despite it is converted to sRGB, using the method described at the bottom of Absolute/Relative Intent, White Point. But beware: you cannot check all values here, since the sRGB conversion compresses out-of-gamut colors, which will change the Lab values correspondingly. Actually, some colors, such as J4 adn L15, are even outside 'Wide Gamut RGB'! I scrutinized the Lab values before any working space or 8 bit conversions — otherwise it won't be correct.

The only patch I found that was slightly off, was F8, which should have been [45.18, -64.70, 23.43] but it was around [45, -62/-63, 22/23], so the a-value was a bit off. Also J4, L15 could be found to be just barely 2 units off in one of the 'a' or 'b' components.


Scarse 0.3, relative colorimetric
it8-image
Click to see a larger version in a separate browser window (136 kB).
This image is scanned in SilverFast Ai at gamma 1.0 with the setting '48 Bit HDR Color', dedusted, assigned a profile produced by Scarse 0.3.

As explained in it8tests.shtml, the profile was made by the author of Scarse from an image of lower resolution (about 1 Mpixel TIFF images) in December 2003. Both age (one year ago) and lower resolution may perhaps cause some deviations. However, I have compared this image with the one-year-old image that was used to create the profile, and when both images are assigned that profile, there are no visible differences at all!

Hence, I conclude that at least the scanner has not changed at all, and since high resolution is not really needed to make a profile (but bit depth must be high enough), I conjecture that the comparison is relevant.

So, even though it may not be 100% accurate to compare this one with the inCamera version, the test supports the conclusion that the differences I see between Scarse and inCamera are significant, and are basically not caused by a difference in resolution of the scan used to produce the profile. So, I make the following claims:

Overall, it looks similar to the inCamera version. But the most striking difference in visual appearance (looking at both images in Lab mode before sRGB conversion) is that the whole yellow column 15 is much less saturated in the Scarse version. Also, the pale patches A13–A19 are almost completely washed-out! (The visual appearance is here about the same before as after conversion to sRGB.) Patches B8, D12 and G22 are very different between inCamera and Scarse, and the whole area A2–L4 clearly change appearance when you swap the two windows back and forth.

Scarse 0.3, absolute colorimetric
it8-image
Click to see a larger version in a separate browser window (132 kB).
Just as is done with the inCamera version, this image is converted so as to maintain the Lab values despite it is converted to sRGB, using the method described at the bottom of Absolute/Relative Intent, White Point. And remember: you cannot check all values here, since the sRGB conversion compresses out-of-gamut colors, which will change the Lab values correspondingly. I scrutinized the Lab values before any working space or 8 bit conversions — otherwise it won't be correct.

Patch GS22 has the right L-value, but it is a bit off in the a/b values, and patches GS22 and GS23 has the same L-value, but it should differ 1 unit. So the dark shadow performance has some problems here.

The patches that visually differed relative to the inCamera version (as described above for relative colorimetric) did of course also have Lab values that were off relative to the reference file.

CONCLUSION
InCamera delivered a close-to-perfect profile, as seen and measured on the IT8 target itself. This does not prove it is perfect all-over, since it does not tell us how it behaves "between" or "outside" the patch colors — for that we need to look at "real" images. One such comparison for a particularly difficult image is presented in the following page.

The Scarse profile did pretty well, but it has som problems. It is worth a closer inspection, and it does certainly have potential.

The SilverFast profile is not recommended at all. As long as it is based on the preview of the scan, it does not have any potential. The HDR version of SilverFast, however, may have potential, since it works very differently.

Back to the beginning of the 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