Stretching a Scanner's Dynamic Range??

stevebrot · 08-11-2009, 04:02 PM

This is a continuation of a discussion that started on another thread regarding scanning technique and extracting the maximum available dynamic range from a film negative. While the most direct route would be to buy a Nikon or Imacon dedicated film scanner, the entry ticket is a little steep.

So...The question is thrown out...

How Does One Best Extend the Dynamic Range For a Scanned Image???

Steve

(Mods...feel free to relocate this if deemed appropriate...)

Gooshin · 08-11-2009, 04:09 PM

i think it was Wheatfield, that posted once, about using multiple passes for, essentially, multiple exposures, almost akin to HDR.

stevebrot · 08-11-2009, 04:20 PM

That is the kind of approach I was wondering about, though I would suspect that image registration would be a problem. I will drop him a PM and see if he has any wisdom in this area.

Steve

Gooshin · 08-11-2009, 04:33 PM

from your other thread

"Strange as it may seem, not all scanners are created equal in this department. The Nikon 5000 ED has a deliverable density range of 4.8. This is 30% better than the Plustek 7500i and 20% better than the Epson V700."

as always, the devil is in the details.

apperantly the 9000ED is an even more capable machine.

my friend (Mischivo) got to use the one at Ryerson for a bit, its a nice piece of technology for sure. (i'm actually trying to convince him to sell off his epson, "our" 5000ED, add a few more bucks, and pick up the 9000ED so that we could do MF as well...)

stevebrot · 08-11-2009, 04:52 PM

The devil is indeed in the details. The high DR is one reason I bought the Nikon 5000 ED. However, that is not a good option for most people. Way too expensive...

I am interested trying some sort of HDR approach, I think I have a couple of negatives that are full range.

Steve

BTW...I don't think you would like the 9000 ED 35mm film holder...

*lbenac* · 08-11-2009, 05:52 PM

Originally Posted by stevebrot

That is the kind of approach I was wondering about, though I would suspect that image registration would be a problem. I will drop him a PM and see if he has any wisdom in this area.

Steve

That is what I read around (Photo.net,...).
You loose in sharpness and do not really gain much in DR.
When I tried it I did not see much difference either way but it might be that none of the negative I used could gain from it.

As always your results might vary.

Cheers,

Luc

Nesster · 08-11-2009, 06:32 PM

SilverFast: Scanner Software, Printer Software and Software for Digital Camera and Imaging :: LaserSoft Imaging
Scanning Basics 101 - All about digital images -- there are a few bits on dynamic range and other useful info
VueScan Scanning Software

Both silverfast and vuescan have multi-pass dynamic range extending modes, and silverfast talks about hdr type of stuff as well. I can't find it tonight, but back when I was doing a bit of research I saw some measurement of how these programs were able to extend the DMAX of various scanners by some not insignificant amount.

I 'use' vuescan, though it's too difficult a program for me (and I'm a programmer)... but there is a way to maximize the range - you set the film base exposure and color, then find white point etc, and do multiple passes too (this opens up shadows), and a dozen or so minutes later, assuming the crop actually was what you thought it was... voila, a really good scan.

Gooshin · 08-11-2009, 06:53 PM

one thing i would like to point out thought, is that this is not necessary from every scan.

if a single pass can pick up 95%, that extra pass may not be needed. (and boy is it timing consuming)

the shot in the other thread, with the two towers shot out of the stone window, that was just a single pass with some shadow enchantment using Lightroom.

stevebrot · 08-11-2009, 07:26 PM

The Silverfast feature is pretty impressive from the examples I have seen. I may be remembering this wrong, but I isn't the multi-pass feature of Silverfast limited to certain versions? I also seem to remember that they only support certain scanners as well.

Here is a link to their page on the feature:

SilverFast Multi-Exposure :: LaserSoft Imaging

As for some sort of multi-pass using Vuescan...that is a new one on me. Can you provide a link?

Steve

Nesster · 08-11-2009, 07:46 PM

Steve Pearce Photography - The Advantages of Multi-Pass Scanning

the multi pass is hidden somewhere in the hendrick website...and somewhere on the www...

stevebrot · 08-11-2009, 07:48 PM

Here is a good article on dynamic range and scanning...

Dynamic range, 24 bit vs 36 bit

*Wheatfield* · 08-11-2009, 10:38 PM

This subject looks pretty well surrounded already. I've had a few occasions when I've wanted to extract more from a negative than I was able to get from a scanner. I didn't take a very scientific approach, I just did 3 scans, one with boosted shadows, one "normal" (but with clipped highlights and shadows) and one with the highlights brought down to a level that showed detail.
This was before the whole HDR thing, so I combined the scans as three layers in Photoshop and played with layer opacity and the eraser tool until I had something I liked.

*edl* · 08-12-2009, 05:33 PM

Thanks Nesster for posting those links.

I found "Scanning 101" to be helpful (go ahead and laugh) so just as a test, I re-scanned a 6x6 Ilford FP4 at 1200dpi, 2400dpi, 3200dpi, and 4800dpi on my Epson V500. The resulting TIFF's ended up around 13, 45, 140, and 340MB's.

ICE was turned off.

Detail continued to improve as the dpi's went up, but what made more difference was properly adjusting the black and white points, and adjusting curves for higher contrast. I have very little PP skill so that wasn't surprising.

Of course, it's a low-end flatbed scanner, and I don't think its 4800dpi output would rival 1200dpi from a real film scanner. But it's good enough for web sharing.

Just my 0.02.

Gooshin · 08-12-2009, 06:34 PM

Originally Posted by edl

Thanks Nesster for posting those links.

I found "Scanning 101" to be helpful (go ahead and laugh) so just as a test, I re-scanned a 6x6 Ilford FP4 at 1200dpi, 2400dpi, 3200dpi, and 4800dpi on my Epson V500. The resulting TIFF's ended up around 13, 45, 140, and 340MB's.

ICE was turned off.

Detail continued to improve as the dpi's went up, but what made more difference was properly adjusting the black and white points, and adjusting curves for higher contrast. I have very little PP skill so that wasn't surprising.

Of course, it's a low-end flatbed scanner, and I don't think its 4800dpi output would rival 1200dpi from a real film scanner. But it's good enough for web sharing.

Just my 0.02.

its not even which scanner is what, its about native DPI, what the scanner was *designed* for

anything outside the native DPI is a combination of software and hardware downsampling, just like shooting a 10mp jpeg using a K20D, when the native resolution is 14 and the output is RAW.

by scanning at native DPI, you are capturign the maximum detail the sensor can produce.

the reason (and i'm theorizing), flatbed scanners have incrased DPI compared to dedicated scanners, is the same idea behind a 10 megapixle point and shoot vs a 6 megapixle DSLR.

however, if your computer has the resources to process 300-400 mb tiff files, then you could down-sample after the fact, since something like photoshop has a much better re-sizing algorithem than Vuescan or whatever.

*edl* · 08-13-2009, 01:07 PM

Originally Posted by Gooshin

its not even which scanner is what, its about native DPI, what the scanner was *designed* for

anything outside the native DPI is a combination of software and hardware downsampling, just like shooting a 10mp jpeg using a K20D, when the native resolution is 14 and the output is RAW.

by scanning at native DPI, you are capturign the maximum detail the sensor can produce.

the reason (and i'm theorizing), flatbed scanners have incrased DPI compared to dedicated scanners, is the same idea behind a 10 megapixle point and shoot vs a 6 megapixle DSLR.

however, if your computer has the resources to process 300-400 mb tiff files, then you could down-sample after the fact, since something like photoshop has a much better re-sizing algorithem than Vuescan or whatever.

Thanks for the explanation Gooshin, that makes sense. I'll give the V500 another go at native res (which I'm assuming is 6400dpi) and then downsize it in PS.

What I'd really like is a scanner where you could just feed the negative strip in and it does everything else (setting black and white, dust removal, etc). I bet those are expensive though. Surely it would cost more than my "point & shoot" V500