LS5000 and 4000 dpi scanning

[tuscland]

Hello everyone,

I am using a Mac Pro and Silverfast 6.6.0r3a Ai (multiexposure).

I am trying to scan an image at 4000 dpi which is the highest resolution available on the LS5000.
Unfortunately I am getting a sort of scaled version of a 2000 dpi scan clearly showing jagging. I have checked all the parameters, and again, but nothing that could solve this problem.

If I use Nikon Scan, every other vertical line is corrupted (note that color scan shows the same symptom with a red hue). Anyway, I don't use Nikon Scan …

v**s**n is the only one that give a decent result at 4000 dpi.

Please someone can help me sort this problem out? Can it be a bug?
I like Silverfast better than the others, and I want to be able to scan at full resolution.

Thank you very much in advance!


Best Regards,
Camille

[LSI_Heidorn]

Dear Camille,
are you 100% ( hard to judge from this crop ) that the SilverFast image simply looks a bit "jaggy" because the Automatic Sharpness is switched on where the v**s**n Image is simply a bit blurry ?!

Anyways i notified testing of this issue !

Greetings,

Nils Heidorn

[tuscland]

Hi Nils,

Keeping the discussion outside of comparisons between different softwares, from the many tests I have made with Silverfast I am 100% sure that it doesn't produce a 4000 dpi image and that this image is made of enlarged pixels. There is no sharpening used here.

Thanks for notifying the testing team.


Best,
Cam

[degrub]

i remember some tests done across a variety of scanners (CS4000 to drum scanners) and the resolving capability of the CS5000 was around 2200-2700 ppi. So any scans at above that should show some artifacts. From the images, it does look like the edge contrast is higher for the Silverfast scan.

[michameijer]

Hi,

I also encountered this problem with my Coolscan 5000ED and SilverFast on Mac OSX, and I'm 100% shure is's not because of sharpening. There's simply information missing in the scan. To me it looks like Silverfast is reading only the half of the image (skipping one line everytime) and later tries to interpolate inbetween. It's even stranger that the problem doesn't appear in Silverfast running on a Windows machine: scans are perfectly fine. I think I read somewhere thas Lasersoft developed their own drivers for Nikon scanners because of Nikon not supporting Leopard. Could this be the problem?

And... can this be fixed? I really like to use Silverfast, but I'm not going to until this is fixed.

Regards,

Micha

[LSI_Morales]

Dear michameijer,
Dear Tucsland,

We conducted many tests to try to reproduce this situation, there are no lines missing in SilverFast scans as suggested by you guys.

There is in fact a difference (as seen in the pictures posted above) probably due to the fact that SilverFast uses a different anti alias algorithm than the other program, the other thing (I am speculating at this point) it seems as v**s**n is using some sort of blurring to make the borders appear softer than in SilverFast or NikonScan.

@ michameijer, we conducted our tests on windows computers as well as macs, the outcomes are the same.

[michameijer]

Dear LSI_Morales,

Would it be useful to upload some pictures or mail them to you (Lasersoft), to illustrate my post? I really want to find out what's going on with the Coolscan/Silverfast combo because Silverfast seems such a nice program to me. Maybe I'm doing something wrong. My tests until now show the best results in Nikon Scan (really, really detailed and sharp, but not 'sharpened' and no artifacts), and average results in v**s**n (lines showing in the shadows of a positive / higlights of a negative). In v**s**n there's an optional 'fine mode' which more or less gives the same results as SilverFast on my mac: 'bigger' pixels and jagged lines.

Regards,

Micha

[LSI_Morales]

Dear Micha,

You can post your pictures here or even better, you can open a ticket on our support system which will be handled by our support and test staff. This will be found by going into our support assistant and and skipping all entries (by answering no to all them) at the end you will be faced with a form which you will fill out and send.

Cheers

[briank]

Does anyone happen to know of any measurement of the true resolution of a Nikon Coolscan 5000 ED? I know that the marketing claims 4000 spi, but I would like a closer to actual number if there is one to avoid interpolation in a high resolution scan.

[LSI_Morales]

Dear Briank,

The scanner sends 4000 dpi as the highest optical resolution without interpolation.

cheers

[degrub]

Are you interested in the maximum sampling or the resolving capability of the scanner ? 4000 spi is correct for the sampling #. i have seen numbers around 2700-2900 ppi for actual resolving capability of the optics + sensor. Have a look here
www.scannerforum.com select comparisons, class 3 - there is a Nikon CS8000 which will be about the same as the 5000.
You can also look here for a more subjective comparison
http://www.terrapinphoto.com/jmdavis/
You can use the approach from this site
http://www.imatest.com/docs/sharpness.html#newtest

to evaluate your individual scanner. ( a slide with a piece of foil mounted supposedly makes a pretty good edge test)

There are also threads in comp.peripherals.scanners archives that have useful information - search for author Kennedy
http://groups.google.com/group/comp.per ... this+group

[hassiman]

Dear Degrub,

What would you recommend as the DPI to set for best results when using Silverfast AI and the Nikon SC 9000?

[hassiman]

Here is a reply from another forum re 8000/9000 resolution:

The first URL the Aztek - Phil Lippincott classification of scanners is
rubbish. The Nikon 8000 should be in a category higher if the categories
mean anything.

The Nikon 8000-9000 always scans at its maximum sampling resolution =
4000 SPI.

There have been German ISO measurements (C'T and ColorFoto) on the Nikon
8000 that described the measured resolution as 3600 PPI. So 10% off on
the manufacturer's number which is very decent.

Or check the results here:
http://www.jamesphotography.ca/scanner_test.html

I measured my Nikon 8000, at MTF50 = 29,6 Cy/mm + 26,3 Cy/mm, at MTF20
138 Cy/mm + 167 Cy/mm.
Target measured in a dry scan without glass was even a bit higher but that
doesn't deliver a completely flat frame for a 6x9 scan.
I'm using my own designed 5 degr. slanted edge target that has a 5,3
Dmax, lith film base.
.

[degrub]

Depends on what you want to do with the scan. If you are enlarging to print, then i scan at 300 ppi X film frame size X enlarging factor and then to the next highest 2x multiple of the sensor ( never above sensor resolution), eg - 10x enlargement of 1 inch frame = 3000 pixels - scan at 4000 ppi and down sample. You could also scan at 2000 ppi and create pixels with photoshop or one of the other programs that does a good job creating pixels. Allowing the scanner over sample at 4000ppi doesn't hurt unless you start to see artifacts in the scan that are not in the film. Otherwise, i scan at a x2 multiple of the scanner elements (4000 ppi, 2000 ppi, 1000ppi etc. In my experience i can scan at 4000 ppi and usually not see significant artifacts. But if i only need 2000 pixels, no point in scanning higher. If i am scanning just to archive, i will scan at the sensor resolution and then 1/2 sensor resolution as a raw file and then use whichever works for the end purpose.

[hassiman]

Dear Degrub,

For some reason I have been having an inordinate amount of trouble grasping the sizing issue vis a vis Silverfast. I was wondering if you might walk me through this process.If I want to make a scan of a 35mm transparancy full-frame for the largest print possible ( with high quality photo output to an Epson 3800 which has a native DPI of 360) with the highest resolution possible using Silverfast AI and the Nikon SuperCoolscan 9000 what settings should I use in the FRAME data ORIINAL-SCALE%, OUTPUT,Q factor, SCREEN, DPI fields?
File size is not an issue.
I am thinking that if I always scan for the largest print possible I can always dowwn size the image file which I would think is better than up-rezing it...?