Scanner bake-off: initial wave of results compiled and published...

[Bart van der Wolf]

Philip Homburg wrote: SNIP

I don't know Bart's slide, I know mine.

Philip was referring to the example 1. at:
http://www.xs4all.nl/~bvdwolf/main/foto/scan/se5400/se5400.htm

A 2700ppi scanner has a Nyquist frequency limit to resolution at 53.15
cycles/mm. When combining/multiplying the MTFs from camera lens (assuming
perfect focus and zero camera shake), film, and scanner lens and sensor, the
actual imaging chain resolution cannot exceed the lowest contributing link's
MTF, and is more commonly around 80% of the theoretical maximum of the
weakest link if all other element are good.

As we have seen with the bakeoff-slide even with careful focusing
there may be tolerances that reduce resolution. And yes, I am saying
that a LS-10 gets 55 lp/mm from a good quality slide.

Sorry that's impossible, unless aliasing makes you misinterpret the results.

Bart

 

[Kennedy McEwen]

I got 118
linepairs/mm with a frontlit 1:8 to 1: 16 selfprinted testchart on
sensia II 100. I did that myself ;-)

Which isn't too different from what Sensia can deliver and well within
the tolerance of your measurement, but still a long way short of your
claimed 200lp/mm for E-100G - roughly half, in fact. Coincidence? I
don't think so!

well, I myself took the E 100 G to 94 linepairs/mm using an old MF
Nikkor 1,4/50 (the first edition of this lens) with a frontlit (1:8 to
1:16) selfprinted target.

Again, about typical for the film's expectations, and incredibly close
to the theoretical calculation I gave you above of 100lp/mm, not
200lp/mm! Another coincidence? I doubt it!

Fotomagazin uses a Leica 2,8/100 macro lens
and a backlit 1:1000 target

And you really think that the difference on the lens will double the
performance? Bunkum - the aerial image of either lens will
significantly exceed any colour film's capabilities. Perhaps if you
referred to fine grain, high resolution black and white tests then you
might have a case, but not with colour emulsion of that published
performance. You are confusing lines and line pairs - or the magazine
article you are quoting from is.

Your
theory may be good, but in real life I get about twice your numbers.

Not according to your own statements you don't - you get about the same
numbers as the theory predicts. Another coincidence? Or validation?

You refer to an unverifiable magazine test which gets twice the numbers,
values which exceed the film's specified capability by about that same
factor, and the same factor that refers l/mm to lp/mm. I believe that
is more than just coincidence, it is a smoking gun!

 

[degrub]

"I count the white lines as well and call an white and a

black line a linepair."

Pardon my ignorance, but i thought a line pair was black and black. That
might be the source of the 2x diffence in the results ???

Frank

 

[WD]

Ralf,

For those of us here in the US, could you be so kind as to
list the resolutions found by the c't magazine?

Thanks so much in advance,

WD

 

[Philip Homburg]

"I count the white lines as well and call an white and a

Pardon my ignorance, but i thought a line pair was black and black. That
might be the source of the 2x diffence in the results ???

No, a line pair is a black and a white line. A cycle is more or less
the same (square wave versus sine wave).

I have to admit, that this stuff is getting quite confusing. My 'real world'
experience sort of matches with the results in the scanner bake-off and
with the results of Bart. These results also matches up with digital
cameras: 6 Mpixel models provide most of what you can do with 35mm,
above 10 Mpixels you are clearly above 35mm.

On the otherhand, Zeiss claim that they can get 200 lp/mm from Ektar 25
in real world pictures (and they have a couple of pictures in
http://www.zeiss.de/C12567A8003B8B6F/EmbedTitelIntern/Lens_Production/$File/Lens_Production.pdf
to demonstrate this). The MTF graphs in the datasheets also suggest the
same if you follow Norman Koren and assume that the film MTF is a second
order filter. On the other hand, the 1000:1 results suggest that you don't
get anything at 200 lp/mm.

However, a 12 Mpixel camera cannot do more than 1414 lp/ph, which translates
to 59 lp/mm on a 35mm frame. If it is that easy to get more than 100 lp/mm
on a frame, it should be easy to get more than 59 lp/mm into a (large)
print. Somehow it doesn't quite work that way.

If the 1.6:1 contrast result for Provia should be interpreted as 60 lp/mm
then I can't understand why a Canon 1Ds has a higher resolution than 35mm
film. On the otherhand, that observation is consistent with 30 lp/mm and
with the 60% or so MTF you get at 30 lp/mm.

 

[David Chien]

Anyways, for these scanners listed on this site, the lp/mm is given in
each set of test image samples reviewed:

http://www.imaging-resource.com/SCAN1.HTM

Gives you a good baseline to work from and a range of lp/mm from various
low to high resolution scanners from the past.

 

[Philip Homburg]

Anyways, for these scanners listed on this site, the lp/mm is given in
each set of test image samples reviewed:

http://www.imaging-resource.com/SCAN1.HTM

Gives you a good baseline to work from and a range of lp/mm from various
low to high resolution scanners from the past.

I don't think that imaging-resource.com is going to help much. First
of all, most the scanners in the top 5 of the bake-off are not present.
Second, both the LS-4000 and the LS-8000 get 57 lp/mm but the bake-off
suggests that there might be a difference. But the biggest problem is
correlating the results from a relatively high contrast target with the
MTF50 results from the bake-off.

 

[Raphael Bustin]

...at http://www.jamesphotography.ca.


pls read my disclaimers before ripping my head off.....

Hello, Jim.

Just discovered the slide you sent.

Don't know how long it was sitting in
the kitchen. Whatever. The scan
was made, JPGd, Zip'd and uploaded
to you .pl address a few moments ago.

Lemme know if there's any problem
with the upload.


rafe b.
http://www.terrapinphoto.com

 

[Bart van der Wolf]

SNIP

I don't think that imaging-resource.com is going to help much.

Although it'll give an impression, they used an approach that is not suited
for discrete sampling systems. The bi-tonal pattern is intended to be used
on analog imaging systems, i.e. camera optics and film. Sampling creates
sensitivity to alignment with the sampling array. To avoid that, sinusoidal
targets, or a slanted edge procedure as described by the ISO, are a better
choice.

First of all, most the scanners in the top 5 of the bake-off are not

present.

Yes, it seems they abandoned the scanner evaluation.

Bart