The simple idea to improve the scanning ability of Silverfas

[dmitri]

1) In 6th version of Silverfast I found the very useful feature - multiscanning (my scanner is Microtek filmscan 35). But looking how this multiscanning is done I thought that there exists better solution than to scan 4,8,16-times the entire picture and then merging it into one (it takes time to move the scanning head, to return it back, and head positioning is not perfect from scan to scan). I think that better solution would be to scan each line 4,8,16 times and then only to move the head to the next line. This would shorten the total time because the scanning head would not return 4,8,16 times from the last line to the first as it is done in the current scanning procedure.

2) It is not very difficult to modify the multiscan procedure so that each pass would be done with different exposure (normal, overexposed and underexposed) and then merging the resulting sequence into the picture that has much more Dmax than the CCD of a scanner can afford. This procedure can easily be done with Adobe photoshop. Just scan 3 pictures, normal, overexposed and underexposed, then merge them adjusting the visibility of layers.

3) It is possible to combine 1st and 2nd procedures. So the procedure of scanning would look like this: scanner moves the head to the first line, then scans it 12 times (normal, overexposed, underexposed, normal, overexposed, underexposed,...), combines this into one image of first line, then moves to the second line, and so on. This wil produce the picture with significantly lower noise and higher Dmax.

[LSI_Flyvbjerg]

Your proposal #1 is already solved for scanners that support multiple scanning in hardware (like Nikon LS2000, LS4000...). Those scanners are scanning each line multiple times, calculate the average and the send us the resulting line. For scanners, that do not support multiple scanning in hardware, we have to proceed like it is for your scanner

Your issue #2 is very interesting. I?ll take it up at our next developer meeting.

Eric.

[President_LSI]

Dear Dmitri,

Thank you for your proposal. We very much appreciate such creative proposals, which have a very good chance (esp. #2) to get implemented quickly.

kind regards

Karl-Heinz Zahorsky

[munchello]

WOW! This proposal (#2) to manipulate the exposure during multipass scanning, if implemented successfully, would be the most important development EVER for Nikon users such as myself.

Extending the dynamic range of the LS-2000 in this way would significantly extend the life of the scanner. I have hesitated to upgrade to the LS4000 because I didn't think it was cost effective. Even with the extended dynamic range and 4000 dpi, I thought that I would still have to do multiple scans and layer them in Photoshop in order to take advantage of the extended range. Curve adjustments in the prescan, even with the LS-4000, leave something to be desired in the way of control/satisfactory results. Combining layers in Photoshop requires converting the scans to 8-bit files.

A good algorithm for combining multiple scans in 48-bit mode at different lamp brightnesses to maximize ALL detail with satisfactory contrast and color fidelity would be a truly revolutionary feature (think of incorporating it into the IT8 calibration module!) It would increase the distance between Silverfast Ai and the competition by lightyears.

PLEASE, PLEASE, PLEASE make this development track a number one priority and keep us thousands of Nikon users posted re: which revision includes this new feature.

And thanks for a great product.

[wolke]

Oh yeah, #2 implementation would save me from buying a replacement for my LS30 too!
Please make it possible!

greetings
Wolfgang