Digital BW, The Print

The results of  ISO 16067 tests (new method) by the German
magazines C't and Color Foto.

Flatbed/filmscanners:
Canon 9900F, Microtek Artixscan 1800F,
Epson 3200, Epson 4870,

Filmscanners:
Microtek Filmscan 3600 Silver, Microtek Artixscan 4000 TF,
Nikon LS 50, Nikon LS 5000, Nikon LS 40, Nikon LS 4000,
Minolta Dimage Scan Dual III, Minolta Dimage Multi Pro.

-----------------------------------------

Dynamic range, Color/B&W, Higher is better:
C9900F-2.6, Mtk1800F-3.2,
E3200-3.2, E4870-3.22/3.30,
Mtk3600S-2.2, Mtk4000TF-3.2,
N50-3.18/2.44, N5000-3.08/2.28, N40-3.3, N4000-3.2,
MinDSDIII-2.9, MinDMP-3.2,

The shift between Color/B&W results with the Nikons 5 could be
the Nikonscan software. Negatives are clipped in that software
though that may have changed in the last version. It is likely
that for color they used slides and for B&W negatives. The tests
are done with the carriers and software provided.

Dmax, Higher is better:
C9900F-3.6, Mtk1800F-3.8,
E3200-3.6, E4870-3.91,
Mtk3600S-2.3, Mtk4000TF-4.4,
N50-4.80, N5000-4.75, N40-4.8, N4000-4.7,
MinDSDIII-3.0, MinDMP-4.3,

Signal/Noise at 2.85 D, Higher is better:
C9900F-0.8, Mtk1800F-5.80,
E3200-3.8, E4870-2.88,
Mtk3600S-0.20, Mtk4000TF-2.0,
N50-1.91, N5000-1.73, N40-4.7, N4000-2.9,
MinDSDIII-1.2, MinDMP-2.4,

The signal/noise ratio at 2.85 D is a new approach in the ISO
test. In the past the Dmax figure was related to the amount of
acceptable noise at a given density. Too much noise meant a lower
Dmax end result then. There are good arguments to do it that way
but this new method is less subjective and makes it easier to
compare results. But at 1.0 and below means unacceptable as the
noise is as much as the signal, 2.85D and above isn't usable
then. For the Mtk3600S with 0.2 at 2.85D the Dmax isn't higher
than 2.3 anyway. For the C9900F with 0.8 at 2.85D and a Dmax of
3.6 it is different. The dynamic range of 2.6 indicates it
already and multi-sampling + longer exposure scans will not help
to reduce noise in that case. That's better with the Epsons.

DOF in mm, Higher is better:
C9900F-2.5, Mtk1800F-1.6,
E3200-2.9, E4870-1.96,
Mtk3600S-0.9, Mtk4000TF-0.8,
N50-0.26, N5000-0.26, N40-0.5, N4000-0.5,
MinDSDIII-1.7, MinDMP-0.9,

Color reproduction in deltaE, Lower is better:
C9900F-11.8, Mtk1800F-9.4,
E3200-6.4, E4870-7.11,
Mtk3600S-9.1, Mtk4000TF-8.2,
N50-6.20, N5000-6.05, N40-6.7, N4000-6.9,
MinDSDIII-10.9, MinDMP-9.4

Lighting evenness in %, Higher is better:
C9900F-97, Mtk1800F-99,
E3200-96, E4870-98.14,
Mtk3600S-98, Mtk4000TF-95,
N50-98.62, N5000-98.94, N40-98, N4000-98,
MinDSDIII-98, MinDMP-98,

The change in the Epson 4870 transparency lighting works.

Length difference to original, fast/slow scan in %, Lower is
better:
E4870-0.00/0.06,
N50-1.00/0.31, N5000-0.50/0.31,
No figures for the other ones.

Pixelshift in pixels, Lower is better:
C9900F-1.5, Mtk1800F-0.7,
E3200-1.1, E4870-0.33,
Mtk3600S-1.0, Mtk4000TF-1.3,
N50-0.16, N5000-0.20, N40-0.9, N4000-1.3,
MinDSDIII-1.6, MinDMP-0.6,

Resolution, Higher is better:
C9900F-1140, Mtk1800F-1320,
E3200-1200, E4870-1650,
Mtk3600S-1950, Mtk4000TF-3180,
N50-3600, N5000-3600, N40-2900, N4000-3850,
MinDSDIII-2880, MinDMP-2840x4960

Middled the slow/fast scan + the two direction resolution results
if they didn't differ significantly.
The Mtk3600S had a focusing problem hence the low result.

Without format choice counted my selection on quality would now
be in this order: S/N, Dmax, Resolution, Lighting evenness,
Dynamic range and then the rest.

Ernst