O
odesmaison
I know this is an eternal question that have come up in the past and I
thouroghly review the archive for 4 days without finding.
First, the subject says it : I'd like to generate custom profiles, so
any indications to best use the scanner (LS-50) and driver is welcome,
but somehow off-topic, so would be recommendations for third part
driver.
Here is what I roughly undertsand (roughly being the key word) :
-the Nikon CMS will always apply (convert) a low-level profile e.g.
NKLS50.icm and NKLS50_P.icm for positive scanning, this would be
assumed profiles. I assume they do not prevent profiling (?).
- in stand alone mode like in twain mode a tag is added in the TIFF but
only in the EXIF, it does not "corrupt" the file. Twain does not pass a
profile anyway (?).
- Auto-exposure does not prevent a color-managed worklow, in the sense
that it maximise the dynamic range of the output by defining the
intensiy at which the CCD wouls saturation, but light being stable and
the scanner self calibrated, it does not impact on hue and saturation.
Just luminance is at stake. Again profiling should (?) be correct under
those circumstances.
- Gamma is set once and for good, and while determine best use of bits
available. I have seen various references-recommandation anywhere from
1.8 to 2.8 (on XP). Ultimately it seems that the best gamma is the one
that provides a match between various patches of the target grayscale
and the reference file measures. This is debatable and advises are
welcome.
Now this would call for simply : turning off NK CMS (with all the
turning off of the driver, the scanner...), keeping auto-exposure on,
determine best gamma and go for the target scan and actual scans. Well,
this gives me below enthusiastic results with a profile generated by a
Monaco commercial package.
At reading knowlegdable people (D. Hitcheson, B Fraser, A Rodney...),
there is much than this to properly profile a scanner :
1. Black point and white point (Dmin-Dmax) need to be determine and
set.
2. While the scanner self calibrates, this does not necessarily provide
a grey balanced response, and best TRC (and gammaTRC) needs to be set
for each R,G,B.
So my questions are 1) am I correct is the above assertions ; 2) how
can I Dmin-Dmax together with TCRs to profile the scanner. I have a
feeling I just ... can't.
Olivier
thouroghly review the archive for 4 days without finding.
First, the subject says it : I'd like to generate custom profiles, so
any indications to best use the scanner (LS-50) and driver is welcome,
but somehow off-topic, so would be recommendations for third part
driver.
Here is what I roughly undertsand (roughly being the key word) :
-the Nikon CMS will always apply (convert) a low-level profile e.g.
NKLS50.icm and NKLS50_P.icm for positive scanning, this would be
assumed profiles. I assume they do not prevent profiling (?).
- in stand alone mode like in twain mode a tag is added in the TIFF but
only in the EXIF, it does not "corrupt" the file. Twain does not pass a
profile anyway (?).
- Auto-exposure does not prevent a color-managed worklow, in the sense
that it maximise the dynamic range of the output by defining the
intensiy at which the CCD wouls saturation, but light being stable and
the scanner self calibrated, it does not impact on hue and saturation.
Just luminance is at stake. Again profiling should (?) be correct under
those circumstances.
- Gamma is set once and for good, and while determine best use of bits
available. I have seen various references-recommandation anywhere from
1.8 to 2.8 (on XP). Ultimately it seems that the best gamma is the one
that provides a match between various patches of the target grayscale
and the reference file measures. This is debatable and advises are
welcome.
Now this would call for simply : turning off NK CMS (with all the
turning off of the driver, the scanner...), keeping auto-exposure on,
determine best gamma and go for the target scan and actual scans. Well,
this gives me below enthusiastic results with a profile generated by a
Monaco commercial package.
At reading knowlegdable people (D. Hitcheson, B Fraser, A Rodney...),
there is much than this to properly profile a scanner :
1. Black point and white point (Dmin-Dmax) need to be determine and
set.
2. While the scanner self calibrates, this does not necessarily provide
a grey balanced response, and best TRC (and gammaTRC) needs to be set
for each R,G,B.
So my questions are 1) am I correct is the above assertions ; 2) how
can I Dmin-Dmax together with TCRs to profile the scanner. I have a
feeling I just ... can't.
Olivier