In the following discussion I compare IT8 targets from three
different sources (Kodak, Coloraid.de and FujiFilm).
These targets cost from around $40 to one costing significantly more (read
hundreds of $'s). Clearly with such a wide variance in cost we would
expect to see significant differences in the quality of the final ICC
profile. However, expectation and reality don't always behave in the way
we would anticipate.
IT8 targets come in two forms: transmissive (IT8.7/1) and reflective
(IT8.7/2), and are commonly used to calibrate film and reflective type
scanners. They are available in three different sizes and
can be purchased from any number of different sources. That said there are
only a few companies who actually produce them. The targets themselves are
based upon an internationally agreed standard which requires that they provide the following
characteristics :
- A set of colour patches that fall within most
device-colour gamut's.
- A neutral "lightness" scale comprising 24 patches ranging from white
through to black. The first and last patches denoting the Dmin and Dmax
values of the target.
- A series of Target-specific colour patches that have been selected
to define the gamut limit of the colorants used to produce the target
- The target will also contain vendor-optional material such as additional
colour/neutral patches or images.
Generic IT8 Target
Apart from cost the main difference between the three IT8 targets that
I'm comparing here is the manner in which they and the associated reference
data file were produced. The Kodak target is mass produced and even targets
with the same batch number seem show a degree of variation. Strictly
speaking the differences between targets within the same batch and the
reference should be within a certain error limit (CIE Lab dE 5) and the
typical error must be reported by the manufacturer for each patch of the
target. The reference data files for Coloraid.de and FujiFilm targets compare
very favourably with the standard. In Table 1 we see the aim values for the
neutral scale lying along the bottom of the IT8 target. The neutral scale
for transmissive targets should have the following L* aim values, based on
the measurement conditions defined in the standard, read from left to right
across the target:
| GS1 |
GS2 |
GS3 |
GS4 |
GS5 |
GS6 |
GS7 |
GS8 |
GS9 |
GS10 |
GS11 |
GS12 |
GS13 |
GS14 |
GS15 |
GS16 |
GS17 |
GS18 |
GS19 |
GS20 |
GS21 |
GS22 |
| 82 |
78 |
74 |
70 |
66 |
62 |
58 |
54 |
50 |
46 |
42 |
38 |
34 |
30 |
26 |
22 |
18 |
14 |
10 |
6 |
4 |
2 |
Table 1 - Lab aim values for
grayscale as defined in the IT8 7/1 standard
Table 2 below shows the batch average grayscale values for the three IT8
targets being reviewed here.
| |
GS1 |
GS2 |
GS3 |
GS4 |
GS5 |
GS6 |
GS7 |
GS8 |
GS9 |
GS10 |
GS11 |
GS12 |
GS13 |
GS14 |
GS15 |
GS16 |
GS17 |
GS18 |
GS19 |
GS20 |
GS21 |
GS22 |
| Kodak |
81.42 |
76.94 |
73.26 |
68.78 |
64.22 |
59.89 |
55.68 |
51.98 |
48.08 |
44.72 |
39.95 |
35.73 |
31.75 |
27.16 |
23.59 |
19.59 |
15.70 |
11.61 |
7.48 |
4.74 |
3.08 |
1.76 |
| Coloraid |
82.39 |
78.69 |
74.68 |
70.76 |
66.72 |
62.91 |
58.84 |
54.77 |
50.81 |
46.32 |
42.39 |
38.60 |
34.55 |
30.59 |
26.42 |
22.40 |
18.35 |
14.32 |
10.07 |
6.52 |
4.38 |
2.02 |
| FujiFilm |
81.43 |
77.50 |
73.50 |
69.41 |
65.34 |
61.74 |
57.59 |
53.63 |
49.71 |
44.79 |
41.98 |
37.42 |
33.99 |
29.88 |
25.85 |
21.72 |
17.79 |
14.19 |
9.96 |
5.67 |
3.68 |
1.65 |
Table 2 - Batch average Lab values
for grayscale
The density of the targets Dmin and Dmax patches will also have
a significant bearing on the targets ability to enable the production of an
accurate ICC profile. If the targets Dmin has a luminance of say 85 (we
really want something in the order of 90 plus) and we scan an image with
brighter highlights (e.g. luminance of 90) these highlight details will be
clipped to maximum white. Good profiling applications will usually allow a
very
small amount of tolerance knowing that making the IT8 targets is not easy,
but there are limits. The goal is to make sure that the Dmin/Dmax of the IT8
target really reflects the maximum possible density range that users will
experience when the film is developed. In order to achieve this the IT8
vendor needs to ensure that their targets actually do include the
maximum density range and that the targets are processed to the highest
standards.
Another area of scanner profiling that causes confusion is that of the
film characteristics. Many folk are of the view that because each film
will tend to have a different colour bias then we need to build a profile
for that type of the film. However, whilst images will tend to vary from
shot to shot depending upon exposure and ambient lighting the film's
colour bias will within certain limits (the exception being reciprocity
law failure with long exposures) remain fairly constant. So in effect it's
not the film that we're profiling, but the scanner response to it. Again,
the technical information provided by Kodak has the following to
say on this matter:
"It is important to note that any colour photographic product contains
both image capture (sensitizing) dyes as well as image-forming dyes. While
the image sensitizing dyes are a vital part of the film performance, it is
only the image forming dyes that affect scanner calibration or
characterisation. Fortunately, only a small number of image forming dyes
are used. Specifically, one set of image forming dyes can be used for
EKTACHROME film, and one for EKTACOLOR paper. This means that only one
Q-60 target is required to represent the complete EKTACHROME film family
and one for EKTACOLOR paper"
Nevertheless, when I set about this comparison of IT8's the jury was still out on
whether we should always match a Fuji target to Fuji films and a Kodak
Ektachrome target with Kodak Ektachrome films, etc. To be honest I've not
yet come across a real world situation where scanning an image captured on
Fuji film whilst using an ICC profile based on a Kodak Ektachrome target
has caused me any great problems; at least not for anything other than precise
colour matching. However, the same cannot be said of Kodachrome type film
as it really does require a matched IT8 target. Likewise it's very likely
that we'll need different targets for Velvia RVP 50 and the new Velvia
100F as these two films are in many respects very different from
each other and the rest of the Fuji range.
The following 3D gamut plot compares the gamut of Kodak Ektachrome and
FujiFilm Provia and should give the viewer some indication of the
differences between these two films. The solid orange area defines the
gamut of the Provia and the yellow "wire frame" area the Kodak Ektachrome.
3D Gamut Plot - Kodak Ektachrome
Vs Fuji Provia
The first step in building a scanner profile is to capture the IT8
target as an RGB image using your preferred scanning application. This RGB
image is then imported into colour-management software that has been
specially written to read the target and compare the values of each patch
with a reference data file. The reference data file will usually be in
text format and comprises rows and columns of Lab values for each patch
making up the IT8 target. By comparing the Lab values of the scanned
patches with the known Lab values of the patches on the target itself, the
software calculates the scanner colour error for each patch. The software
then uses this information to build a colour table that describes how the
scanner reproduces the colour and neutral patches. This table along with
its supporting information is then saved as an ICC colour profile.
As mentioned above, depending upon the type, size and source of the
actual IT8 calibration target the reference data file will either be
unique to an individually measured target or a batch of targets. Clearly
an individually measured IT8 will be more accurate than batch produced
targets but this accuracy comes at greater cost (sometimes significantly
greater). The question is - does the additional cost of a uniquely
measured IT8 target really convert into more accurate colour
matching?
The three IT8 target slides that I compare are:
- Kodak Ektachrome Q60 supplied by Kodak
- Fuji Velvia supplied to me by Wolf Faust at
Coloraid.de
- Fuji Provia supplied to me by
Fuji Germany (no
sales)
The following series of scans were captured by my Nikon LS-4000ED using
LaserSoft SilverFast 6 with the gamma-gradation value set to 2.4 rather
than my usual gamma value of 2.2 Using the higher gamma ensured that
shadows were open and that the final scan is relatively low in contrast.
The actual ICC profiles were created using
GretagMacbeth
ProfileMaker Pro 4.1.5
Clicking each of the thumbnails will open a page that contains before and
after animated comparisons of the IT8 target and a sample real world image. The
sample image (Fuji Sensia II 100 ISO) contains saturated red and blue.
Please note that the animations are NOT intended to demonstrate that one
target is better than another but to show the quite significant
improvements that can be achieved by using a colour managed workflow when
scanning.
 |
 |
 |
|
Click image for - Ektachrome |
Click image for - Fuji Velvia |
Click image for - Fuji Provia |
Comparing the batch average data for each target it is clear that
those from Coloraid.de and FujiFilm are much closer to the
specification than the Kodak target. Actually, other than patch GS22 the
FujiFilm target values are very close to the standard. Whether my
sample targets are better or worse than the batch average I can't say but
my four Kodak targets certainly do vary from each other.
As mentioned above the FujiFilm Provia target was supplied to me
by Fuji Germany as a sample of their IT8 targets. I'm advised that it was
created for FujiFilm for quality assessment and is one of a batch
of 12 targets created from a single sheet of Fuji Provia film.
It's the closest I'm likely to come to an individually measured IT8 target
for a 35mm film. Yet
there appears little to choose between it and the Fuji based Coloraid.de
target in terms of how each reproduces the sample scans. Given the cost
differential between the two I'll admit that it was quite a surprise to
find the final scans from the Coloraid.de
target apparently matching those from the FujiFilm target. So, at around $40
including
shipping it certainly looks like a no-brainer as far as which target
supplier
offers the better cost/benefit ratio. On the other hand none of my Kodak targets
are quite as good as the Coloraid.de target and I would be hard
pressed to recommend anyone spending $'s on one. That said the Kodak still offers
substantial improvements over using no profile at all.
It's worth mentioning that Wolf Faust is of the view that
matching your film and target is important and recommends that Kodak
Ektachrome film users do purchase his Ektachrome targets in preference to the
Fuji targets, and Fuji targets for Fuji film. Given that the test image
was shot on Fuji Sensia II this may well explain my preference for the
Fuji based targets, and effectively contradicts my earlier view that
matching targets to film stock isn't really necessary. Further details of IT8 targets based on both Kodak and Fuji film
can be obtained from Wolf Faust at
Coloraid.de.
September 2003
Another target that appears to have gathered a following is the HCT
target by Don Hutcheson of
HutchColor.
Don has also produced a very informative guide to scanning to
complement his HCT target. Shortly after publishing this article
Don kindly sent me one of his HCT targets. Whilst the 35mm versions
are not true hand measured targets they are still very accurate. This
accuracy is achieved by including approximately twenty-five 35mm and two
larger format targets on a single 10 by 8 inch sheet of film. The larger
targets are then measured and averaged so as to obtain a very accurate
batch-average. At $155 the HCT targets are expensive but they definitely produce very high quality
results. Unfortunately they don't follow the normal IT8 format and because
of this not all profiling applications can read the reference data file.
So you should check that your profiling application is compatible with the
HCT format before ordering one.
HutchColor (HCT) Target
