summaryrefslogtreecommitdiff
path: root/doc/evalInputTargets.html
blob: 91a38f1359520203dd7bf6b01d7b5b9e5615bd89 (plain)
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
<!DOCTYPE HTML PUBLIC "-//W3C//DTD HTML 4.01 Transitional//EN">
<html>
<head>
  <title>Evauating Input Targets</title>
  <meta http-equiv="content-type"
 content="text/html; charset=ISO-8859-1">
  <meta content="Graeme Gill" name="author">
</head>
<body>
<h2 style="text-decoration: underline; font-weight: bold;">Evaluating
input targets<br>
</h2>
There are a variety of color profiling targets available for
characterizing input targets, and it is sometimes hard to decide which
one will be best for your particular purpose. Listed here are some
criteria to evaluate:<br>
<h4>Color gamut</h4>
The resulting profile will only be accurate between or near the color
values contained on the chart. This means that if the colors you are
actually processing with your device go outside the gamut of your test
chart, the color values will have been extrapolated by the profile, and
are therefore likely to be not very accurate. <br>
<h4>Color Resolution</h4>
The profile will be most accurate for colors that are near those
contained on the chart. This means that the more closely and evenly
spaced within the color gamut the value of<br>
the test chart are, the more accurate overall the chart will be. So
typically the greater the number of test values, the better.<br>
<h4>Dynamic range and White point</h4>
Similar to color gamut, the profile will only be accurate over the
range of lightness levels exercised by the chart. At the dark end the
ideal black test value will be a light trap. At the white end the ideal
white test value would be the perfect 100% reflective diffuser. In
practice there is another consideration, which is that by default the
white point of the profile is set by the white value of the test chart,
and any values over this may be clipped by the profile. So ideally the
white patches should represent the white value of the work you will be
using the input device for.<br>
<h4>Spectral similarity</h4>
One of the fundamental problems with colorimetrically characterizing
input devices, is that typically input devices don't have the same
spectral sensitivity as a human observer. This means that it "sees"
color differently, and that there is no way to perfectly compensate for
this in a device profile. [There will be some spectral values that look
the same to the device but appear different to us, and visa-versa.] A
colorimetric profile will best compensate for such differences when the
target test colors have the same spectral reflectance characteristics
as the the intended work. What this translates to is that you will get
best results when the test chart uses a similar printing process to
whatever work you will be using the input device for. So if you are
intending to scan photographic prints, you should use a photographic
based test chart. If you were scanning artworks, then you should use a
test chart that has pigments that are similar to paint used on such
artworks, etc. When characterizing camera's, an additional source of
spectral differences is the illuminating light source used. Once again,
it will be best to choose a light source to characterize the camera
that is going to be most similar to the light source you will typically
shoot photographs under. <br>
<br>
</body>
</html>