How much should I be expecting from my scan?

[Mark F]

I've never been able to get really good scans from my Nikon 5000ED (using Vue Scan) and I'm wondering if I'm just expecting too much. Still, I've seen large 20 x 30 prints made using the same equipment so I know it can be done.  I've attached an image as an example. The original is a Kodachrome 64 slide. On the original the red leaves are clearly reflected in the pond and the brownish leaves on the trees at the upper right are also reflected, much more dimly, but there. On the original the two tree trunks at the left still have detail while on the scan they are totally blown out. I tried increasing the detail in the water by pulling the black point all the way to the left. There is a little more detail but the whole image gets too bright. The scan is much more contrasty than the original.

I've brought the image into both LR3 and CS4 but cannot improve the quality. Am I expecting too much?

For the record, the resolution has been set to 4000 dpi, media to Kodachrome, color balance to neutral and multi-exposure is turned on.

Comments would be appreciated.

[feppe]

I only shoot and scan slides, but I've heard negs are tough to scan, and especially Kodachromes are notoriously difficult. Silverfast has a specific Kodachrome setting to try to overcome that, so you might want to check out their demo. If you decide to buy it, take your time in determining which edition you actually need - their editions and upgrade paths are even more convoluted than Adobe's.

The sample you have might also benefit from scanning it twice: once for highlights, once for shadows, and combining in post by digital blending.

[Mark D Segal]

Your scanner is excellent, I have one too, and I get fine results from it. I mainly use SilverFast because it is a far more refined application in many respects, but you should be getting very usable results from Vuescan, much better than what you are showing here. The main thing you need to be really careful about is not clipping the histogram at either end of the scale. Once you have all the data in an image file unclipped, you have choices about how to handle it. As a first-round diagnostic procedure, use the so-called "raw" scanning option, which effectively just picks up all the information from the scanner without making any edits in the scanning application, save it to your hard-drive and open the image in Lightroom. Before you do this scan, make sure you have Vuescan's colour management set correctly with the right profile for a Kodachrome slide. Also, I recommend an ARGB(98) colour space, for safety, (funny things can happen in scans with ProPhoto), turn off multi-exposure if the raw option doesn't turn it off automatically, and save in the TIFF file format, not JPEG. Once you open that scan in LR, the histogram should not be clipped, and if anything the image could perhaps look a bit dull and flat if all your LR settings are set to zero and linear. That gives you a good base from which to build the image luminosity and colour the way you want it. Once you've tried all that, please let us know what you observe.

[Mark D Segal]

I only shoot and scan slides, but I've heard negs are tough to scan, and especially Kodachromes are notoriously difficult. Silverfast has a specific Kodachrome setting to try to overcome that, so you might want to check out their demo. If you decide to buy it, take your time in determining which edition you actually need - their editions and upgrade paths are even more convoluted than Adobe's.

The sample you have might also benefit from scanning it twice: once for highlights, once for shadows, and combining in post by digital blending.

Hi Harri,

He's talking Kodachrome, and he shouldn't be having the problems he's showing, even using Vuescan. I think there's something messed-up in his settings. Your idea that he should download and try a demo of SilverFast is a good one, but I think he may be able to solve his issue with what he has - at least to the point of getting an information-complete scan into Lightroom for the heavy-lifting. If he can't, that was going to be my next suggestion. And yes, deciding which version of SilverFast one should get is not for the feint of heart, but in the final analysis, the key break-points turn around whether or not one wants just a few basic features which really separate all those versions from each other. We can get to that if need be. :-)

[rgs]

I only shoot and scan slides, but I've heard negs are tough to scan, and especially Kodachromes are notoriously difficult. Silverfast

Kodachrome is slide film, "chrome".

[jaapb]

For the record, the resolution has been set to 4000 dpi, media to Kodachrome, color balance to neutral and multi-exposure is turned on.

Mark,

What is your output and output settings, Vuescan "raw" Tiffs are very dark (gamma 1.0) and are supposed to be an archive for further "scanning from file" in Vuescan.
Also consider to set Crop|Buffer to say 10%. This causes Vuescan to exclude 10% of the outer scan surface when calculating colorbalance. When you scan maximum surface, chances are that white or black film-edges are included in the scan and mess up calculation.
I too have a 5000 and it is a very good scanner and you should be able to get good results provided that your orginal is not too bad. Scanning is not always straightforward and usually an elaborate process.

Also, I recommend an ARGB(98) colour space, for safety, (funny things can happen in scans with ProPhoto)

Mark,

What is wrong with ProPhoto when saving out scans. It is my workflow now to save ProPhoto tagged Tiffs and I can't really see anything wrong with it, although I haven't compared it to ARGB.

Jaap

[Mark D Segal]

Mark,

What is wrong with ProPhoto when saving out scans. It is my workflow now to save ProPhoto tagged Tiffs and I can't really see anything wrong with it, although I haven't compared it to ARGB.

Jaap

There's nothing necessarily and systematically wrong with it, especially for positives. For negatives I've seen the risk of exaggerated reds and a slight warm imbalance to the whole image. The reason why I recommended he stick with ARGB(98) for now is that he's in diagnostic mode, and I think it best to limit the variables that can potentially introduce issues.

[Mark F]

Thanks to all for helping and especially to Mark. As the attached revised scan shows there is a substantial improvement although the image obviously still needs work, especially in the reflections. When I brought the image into LR3 and burned in the lower right water I could see that there is some detail of the reflected red bush and tree leaves but much less than shows in the slide. It is a difficult burn to do as it looks manipulated and I am wondering if there is a way to get even more detail in the scan? There is no clipping on the black end. The input bits per pixel is 48 bit RGB and I am doing 4 samples. Is that enough? 

Thanks again.



 

[Oldfox]

I've never been able to get really good scans from my Nikon 5000ED (using Vue Scan) and I'm wondering if I'm just expecting too much.

Strange, I have used Nikon Coolscan V ED ("little brother" of 5000ED) with NikonScan 4 software and have seemingly better results. The attached sample has been scanned with 4000dpi, Kodachrome and little lightening in PS, nothing else. The other sample is warmed in PS, the original seems too cool to me.

Have you any other samples? Maybe people or a streetview in the picture?

/fox

ps. I didnt see your second sample while writing...

[Mark D Segal]

Thanks to all for helping and especially to Mark. As the attached revised scan shows there is a substantial improvement although the image obviously still needs work, especially in the reflections. When I brought the image into LR3 and burned in the lower right water I could see that there is some detail of the reflected red bush and tree leaves but much less than shows in the slide. It is a difficult burn to do as it looks manipulated and I am wondering if there is a way to get even more detail in the scan? There is no clipping on the black end. The input bits per pixel is 48 bit RGB and I am doing 4 samples. Is that enough? 

Thanks again.
 

You are welcome, but what did you actually do to get this much improved result? Need to know that before suggesting other measures! And yes, four samples is (more than) enough.

[Mark F]

Mark, I followed your original post and the changes you suggested. I'd bet that what helped most was the elimination of clipping of the highlights. Your comment that my settings must be screwed up was on point. Under the Preferences tab, even though I had the histogram set at linear, the graph type was set at "curve" so I was not getting a true histogram and did not detect the clipping.

Oldfox, I've found that I can usually only see images if I am signed in. Of course you had to be signed in to post so I do not understand why you were unable to see my posted image.

[Oldfox]

Look at the timestamps of the posts. You submitted your post while I was selecting my examples and writing my post.
/fox

[Bart_van_der_Wolf]

Mark, I followed your original post and the changes you suggested. I'd bet that what helped most was the elimination of clipping of the highlights. Your comment that my settings must be screwed up was on point. Under the Preferences tab, even though I had the histogram set at linear, the graph type was set at "curve" so I was not getting a true histogram and did not detect the clipping.

To avoid clipping you have several options for visualization of it available in VueScan. You have color options to show certain (user definable) colors for certain events (such as clipping, or IR data if applicable) as an overlay on the preview, and you can set the histogram to show the Raw data, and you can change the vertical scale of the histogram to e.g. Logarithmic to allow critical judgement of low histogram bin counts. That should alow to tweak the exposure for optimal scan quality.

Cheers,
Bart

[Mark D Segal]

Thanks to all for helping and especially to Mark. As the attached revised scan shows there is a substantial improvement although the image obviously still needs work, especially in the reflections. When I brought the image into LR3 and burned in the lower right water I could see that there is some detail of the reflected red bush and tree leaves but much less than shows in the slide. It is a difficult burn to do as it looks manipulated and I am wondering if there is a way to get even more detail in the scan? There is no clipping on the black end. The input bits per pixel is 48 bit RGB and I am doing 4 samples. Is that enough? 

Thanks again.
 

Could you post a version of that image before you burned-in anything? Burning-in can clobber detail, but finessing it with Fill and Blacks and other stuff in LR can do wonders for it. If you repost without the burning, I'll re-try what I describe below.

There is no scanning software I know of - not Vuescan, not SilverFast, which offers the kind of refined control over luminosity and colour than you can achieve in LR, provided the original scan fed into it is "in the ball-park" (i.e. no clipping, no massive colour casts, no large areas really scraping the bottom of the tone curve). So I wouldn't try getting the last ounce of image quality out of the scanning software. Use it to optimize the photo for the finishing work in LR and/or PS.

I downloaded your posted image and opened it in LR3.3, then made a few adjustments which I think improved it. So try this in your system and see what it looks like on your display:

White balance: temp +2; Tint -7
Fill Light 44
Blacks 18
Clarity 2
Vibrance 25

[Mark F]

I'm attaching the latest iteration of this scene. There is now a hint of the red leaves in the reflection but not enough to reflect what I saw and what Kodachrome captured.

Mark, the previous post was a version without burning. The "burned" version looked so awful that I discarded it.  Thanks for posting your LR3 settings. That version is a different interpretation than mine, with a darker mood. Interesting.  I lightened my version in order to see some details in the water and also bumped up the Vibrance to make the red leaves stand out.

Bart, thanks for the tip on clipping warnings.  I actually did go through the manual but either my reading comprehension skills have declined or the manual was written for engineers 

[Christoph C. Feldhaim]

Scanning is fun - once you have gone through all the pain ....

[Mark D Segal]

Mark, the previous post was a version without burning. The "burned" version looked so awful that I discarded it.  Thanks for posting your LR3 settings. That version is a different interpretation than mine, with a darker mood. Interesting.  I lightened my version in order to see some details in the water and also bumped up the Vibrance to make the red leaves stand out.

Could be a difference of monitor settings, but my version on my display turned out more vibrant and somewhat brighter. Was your tone curve in LR set to Linear and everything else zeroed? Mine was - I forgot to mention that. You may wish to try a rescan with the lower quarter-tones bumped up a bit in the scan. It may give you a better starting point for achieving more luminance from the leaves in the reflection when it gets into LR. In principle you should be able to extract any colour and luminance the slide can yield, especially with a Nikon 5000. The dense end of slides can be stubborn to anti-up what they really contain, so some tweaking in the scanning software looks indicated for this.

Christophe - :-), yes - if fun means "challenging" no doubt about it!

[feppe]

Kodachrome is slide film, "chrome".

Of course, *facepalm* My excuse is that it's confusing since the product is named Kodachr... umm. Never mind. Nothing to see here. Move along.

[Nick Walker]

Mark,

What gamma setting are you using, 2.2? I wold recommend trying higher gamma settings with high contrast slides - up to 2.8 max. It is image specific but it can help matters. Higher scanner gamma settings will lift the shadows without clipping between the middle tones and highlights.

To save me explaining - http://www.hutchcolor.com/PDF/Scanning_Guide.pdf

It is by trial and error and not always a recommended route.

[Mark F]

Mark, that did the trick. I now have a scan that shows the detail in the reflections. To close this off I am attaching one last version to show the reflected leaves. This is still a work in process as I have lost the blue sky but that should not be too big a problem to correct.  As has been mentioned on another thread, if you ever do a podcast on scanning I would certainly subscribe if for no other reason than to see the differences between Vue Scan and Silver Fast.

Nick, thanks for the reference. I do not know how gamma works but will read that article.