Kodachrome 64 scanning

[miklosphoto]

I have a question about scanning 35mm Kodachrome 64.
I use a Nikon 5000 and Silverfast (latest) to scan my film. My problem is that slides with shadows scan very poorly. If I look at the slide on a light-table I can clearly see details in the shadows. When i scan the same slide, all those details are gone as they are underexposed by the scanner. By the way I calibrated the scanner with Kodachrome target, so that should be fine.
Any tips how to retain those details in the scan? I tipically scan at 2400 dpi and save as tif.
All suggestions and help is very appreciated.

 

[donbga]

I have a question about scanning 35mm Kodachrome 64.
I use a Nikon 5000 and Silverfast (latest) to scan my film. My problem is that slides with shadows scan very poorly. If I look at the slide on a light-table I can clearly see details in the shadows. When i scan the same slide, all those details are gone as they are underexposed by the scanner. By the way I calibrated the scanner with Kodachrome target, so that should be fine.
Any tips how to retain those details in the scan? I tipically scan at 2400 dpi and save as tif.
All suggestions and help is very appreciated.

Since I'm not familiar with the scanner or software I can only guess that the shadows maybe being clipped or crushed by the software settings.

How does the dynamic range of the Q60 calibration slide look? Are you getting full black to white?

Don Bryant

 

[mesh]

Are you using Ai Studio with the new Kodachrome profiles? I guess you are if you have the target, but if not, I found it made a huge difference over the standard Silverfast software (using an Epson V700) There's no doubt that Kodachrome can be a bit of hassle to scan, but once you have sorted it out, the quality off your 5000 should be superb. You can obviously download a demo of Ai Studio to test.
If you're using ai then obviously you just have to play with a curves adjustment as I am sure you've done...

 

[mesh]

One more thing... have you read this?

 

[glhs116]

Try Nikon Scan (the latest, 4.0.2). Let it autofocus and autoexpose. Specify 16bit and 4x. Save as TIFF. No other options. I'll be surprised if you can't get readable shadows. You may need to tweak curves a bit in your imaging application.

You can also try (according to taste) enabling and adjusting the "DEE" option in Nikon Scan. It's a sort of "shadow lift" at scan time.

Lastly, you can try increasing the so-called "analog gain" master control.

Sam

 

[pellicle]

Hi

indeed Kodachrome is a hard nut for the shadows. I've found that using nikon scan and just carefully setting the levels does quite well.

you may also try the multisampling to assist clearing up of noise.

http://cjeastwd.blogspot.com/2009/10/quick-negative-scan-tutorial.html

 

[glhs116]

Yes, sorry for not being clear. When I said "16bit and 4x" I was referring to 4 times oversampling. To me this seems to be a sweet spot. Not that much slower than no oversampling and not a lot of benefit to be had from the higher values. I use 4x as standard when I scan.

 

[Loris Medici]

Oversampling / effective scanning bith depth

I don't know how correct that is but I remember reading somewhere (reliable - could be in Silverfast or Vuescan documentation, I'm not sure...) that your effective scanning bit depth increase by the binary logarithm (log2) of the oversampling amnt. E.g. 8x oversampling adds effectively log2(8) = 3 bits to your hardware scan bit depth... Any comments???

 

[Tom Kershaw]

I don't know how correct that is but I remember reading somewhere (reliable - could be in Silverfast or Vuescan documentation, I'm not sure...) that your effective scanning bit depth increase by the binary logarithm (log2) of the oversampling amnt. E.g. 8x oversampling adds effectively log2(8) = 3 bits to your hardware scan bit depth... Any comments???

Loris,

I can't comment on the mathematics but my recent attempts to scan Kodachrome 64 using VueScan & the LS-9000 with 4x sampling did not pull as much shadow information as is visible on the physical film. I'll make a second attempt with the NikonScan software at some point.

Tom

 

[pfigen]

A Nikon CCD scanner will never be able to record the shadow detail on Kodachrome. Kodachrome maxes out at around 3.8, and despite what Nikon (and any other CCD maker) claims, their scanner will never see anything above about 3.0 without gobs of noise. The only sensor capable of recording all the shadow detail on Kodachrome is a photomultiplier tube, which, of course, means a true drum scanner. And even then, Kodachrome can tax those pmt's as well. Plus you need to know how to compensate for the very different way scanners "see" Kodachrome dyes vs. the way they see E-6 dyes. You can try and find a K-14 IT8 or you can use something like a Hutchcolor target, but you still need to pay close attention to both the shadow and highlight numbers to avoid the blueish/cyan cast that all too many Kodachromes scans exhibit. All scanning software and scanner hardware are not created equal. Kodachrome is always a challenge, not unlike getting great color neg scans, but well worth the effort once you figure it out.

 

[Tony Egan]

I have a rather old Nikon coolscan ED IV and an Epson 4990 flatbed scanner. I was recently scanning Kodachrome (25 I assume) slides from the late 60s and the Epson did a better job overall with colour and shadow detail. I tweaked the histogram as required before scanning and also got some additional benefits from the shadow and highlight adjustment in PS.

 

[Tim Gray]

I've been getting pretty good results doing the following on my Coolscan V:
- scan in Vuescan with multi exposure selected. It does a two pass scan at different exposures, one normal one, and then a high exposure one to pick up some extra detail in the shadows. They are then blended together.
- save as a raw linear scan.
- bring it into Photoshop and apply the profile I generated using the Lasersoft Kodachrome target and ArgyllCMS.
- convert to working space.
- tweak the result with a curves adjustment layer. There are two steps here. The first is to boost the black midpoint (a gamma adjustment) to lighten midtones. Otherwise my scans come out too dark looking. The second is not necessary, as at this point, the image looks pretty darn good, but it does help a little bit in achieving a better match to the slide in my opinion. For the most part, this seems to almost always include moving the black point on the blue curve up a little bit from zero. Minor work with the midpoints of each RGB curve some of the time. I'm not why this whole step is necessary when using a profile, but I seem to be getting better scans doing this, so oh well. Probably something to do with my profile.

 

[Diapositivo]

I second using "multiexposure" as already suggested by Tim Gray.

I use VueScan with a Nikon Coolscan 5000 ED. Silverfast should have a similar function, but maybe not all versions of Silverfast have it. In case your version of Silverfast does not have the multiexposure option - which might have a different name in Silverscan - I suggest buying VueScan because that really is an important factor in the final result.

I always scan with "multiexposure" on and 16x multisampling on. That means that the slide is actually read 32 times. You paid your scanner a lot of money, why set for less than the maximum quality it can give you?

As far as I know, NikonScan does not have the "multiexposure" function or, at least, I couldn't find it.

I save scans as "linear DNG" scan (raw) and work from there.

Fabrizio