Ernst Dinkla said:
What should be the right settings in Vuescan for a Nikon LS8000 with
a lith film 5 degr. slanted slope target to test the resolution/MTF
with Imatest ?
Besides a manual focus series (around the Auto focus value), Raw
output (linear gamma), and underexpose* slightly to avoid clipping. A
crop of, say, 250x800 pixels at native resolution will allow to avoid
imperfections like dust.
*If the scanner can't be fooled into underexposure, adjust exposure on
a frame area that's unobstructed by film, or add a neutral density (or
polarizer) filter just before exposure. This may require refocus. One
ultimately wants some headroom and 'footroom' on the scanned result,
to avoid clipping issues in evaluation.
Which choices should be selected in Imatest to measure the test
image ?
Crop area (e.g. near horizontal and near vertical) or full
cropped image,
SFR data options:
Plot:
- cycles per mm (or pixel) for 4000 ppi (assuming that's
actually correct)
Options:
- e.g. Edge pixel profile (only matters for charts)
Settings:
- Gamma=1.0 (needs to be verified with steptablet),
- Y (luminance),
- no standardized sharpening,
- for width and height use 24x36mm equivalent @ 400ppi =
- 3780x5669 pixels (this will make later comparison to Full Frame
dSLRs easier).
- Crop location is filled-in automatically.
Optional parameters are mostly used in cumulative spreadsheet
overviews.
After calculating, Save SFR results:
Output options: Make sure to select Excel .CSV detailed results.
Other options can be selected depending on further use.
I usually output (but don't save) Cycles/mm or Cycles/pixel
graphs when testing a single device, or LW/PH if comparing
different size sensors. When I find the best result (e.g. due to
focus), I rerun on that data file and do save the "figures".
Are there any samples of Imatest scanner tests + graphs around next
to what Imatest/Norman Koren has on its pages ?
Very few, if any, solid scanner tests that I am aware of.
You can have a look at Jim Hutchinson's Scanner Bake-Off results;
<
http://www.jamesphotography.ca/scanner_test.html>
but be warned that not all contributors respected the rules, and gamma
was not quantified, so results where biased due to postprocessing. In
the mean time Imatest has also matured a lot (it was a Beta version in
2004).
Of course you can also benchmark against my DSE-5400 evaluation:
<
http://www.xs4all.nl/~bvdwolf/main/foto/Imatest/SFR_DSE5400_GD.png>
for scanner only (no camera lens+film), or
<
http://www.xs4all.nl/~bvdwolf/main/foto/Imatest/SFR_Graphs.png>
(needs to be updated), which demonstrates how LW/PH allows direct
comparison between different sized sensors (and output magnification
to same size).
Bart
