Epson vs Nikon

[sanmich]

I am trying to figure out why I can't get proper shadows from my CS 5000 when I can get them from my V500.
The resolution of the Nikon is much better, but the shadows look crappy.
I use vuescan, and tried what seems to be any combination of RAW saving/TIFF/with and without Scanhancer, etc..
I am joining two samples. First one is the CS, second the V500.
Any suggestion?

 

[Elektrojänis]

To me it looks like the firs one has more shadow detail and resolution/sharpness. Because of that the grain shows up more (just quessing why it might look crappy to you). It would be trivial to darken the shadows in post processing (or even in scanning software) to match the second one if you want them like that.

I'm not a pro or anything though... Just wondering why you like the shadows better on the second sample.

 

[sanmich]

I should have added that the histograms are very different in the blacks:

The Epson goes down nicely to zero, while the one from Nikon has a peak at the lowest values, which I understand, means that clipping is occuring

 

[gavinlg]

I personally like the look of the coolscan shadows more - they're crisper with no less detail on my monitor...

 

[thegman]

To me the Coolscan looks better, the shadows are a tad darker of course, but for me, look as nice or better. You could try scanning as a positive, or if you've got Vuescan, try multiple passes.

 

[Rob-F]

Another vote for the coolscan. The Epson looks flat by comparison.

 

[sanmich]

OK, last news from the front:
It seems at least part of the problem is Vuescan:
When I scan with Nikonscan, I get what I feel are better shadows, the histograms look nice in the dark areas and using scanhancer doesn't seem to impact the scan at all. I'll try to learn the software (never used it), design a workflow for digital negs, and, hopefully, back again to those 80 scans I need to do...:bang:

 

[aldobonnard]

OK, last news from the front:
It seems at least part of the problem is Vuescan:
When I scan with Nikonscan, I get what I feel are better shadows, the histograms look nice in the dark areas and using scanhancer doesn't seem to impact the scan at all. I'll try to learn the software (never used it), design a workflow for digital negs, and, hopefully, back again to those 80 scans I need to do...:bang:

Not sure that Nikon Scan betters Vuescan, rather the opposite in B&W... I have a long experience of comparison between Nikon Scan and Vuescan and I get better results using VueScan for B&W, in general. In Vuescan, try to use multi exposure along with 8x multi sampling (16x is overkill). When doing a preview, check your histogram values with the "raw" graph, and adjust the analog gains to fill in nicely without clipping the edges. It works for me. If you are still not satisfied, or if the negs are difficult, then from Vuescan export in Raw/DNG and process them in your favorite software.

 

[carlmuck]

Search for "vuescan lock exposure" in the Google. There's a post on flickr showing how to set the scanner exposure to help with just this problem. It may be that the nikonscan software is better at setting the exposure automatically.

 

[sanmich]

Thanks gentlemen

I am scanning RAW in vuescan, saving as DNG, and still, clipping occurs.
I don't know if there is an option in Vuescan to manually adjust the exposure (manual gain, right?), as I understand there is in Nikon scan, but I tried to lock the exposure on a blank area between frames, and without a film in the scanner (on the scanhancer diffuser), and still got some clipping.
I have read that vuescan can be a bit buggy in the sense that there are lots and lots of controls, and it's too easy to make a mess out of them, so I started from scratch (factory options), but it didn't help.
The multi-pass thing may do the trick, but I will also need to check that I can work with the scanhancer. Without it, the results are a nightmare in terms of defects.

 

[Tim Gray]

Vuescan will let you adjust the manual gain on the scanner, but most of the time the setting that it picks automatically is just fine.

In the color tab, set black point and white point to 0%. This affects clipping.

This is more or less how I scan B&W on a CS V with Vuescan. I might have some minor changes now, but it's more or less the same thing:
http://125px.com/articles/photography/digital/vuescan/

As far as the DNG thing, apparently I'm the only person in the world who thinks it's a bad choice, but you can edit TIFFs (and jpegs) in Camera Raw and Lightroom, so I'm not sure what you gain by using a DNG. ACR will even store the ACR settings in the TIFF file too. I'm pretty sure Lightroom would as well.

 

[sanmich]

Tim, thanks a lot for the link,

about the zero settings black and white setting:
I will have to check again, but I'm almost sure that it's what I am using.
At any rate, as I understand it, it's not an adjustable exposure/gain thing but a way to ask the scanner to keep values of black/white or to transform them in pure black/white while allowing for a better contrast in the midtones. Also, it shouldn't impact the raw file, should it?
I was using RAW/DNG exactly for the reason I want as little tweaking made by the scan software as possible. If I start using Nikon scan, I'll have to use TIFF anyway.
Why do you think DNG is a bad idea? do you mean as compared to RAW-TIFF? or to regular, adjusted TIFF?

[edit]: a very interesting thing in your link is the "make grey" from BLUE setting. Any idea why is it set so?
[edit 2]: the article explains about the use of the blue channel...

 

[aldobonnard]

The manual gain in vuescan pilots the nikon scanner's hardware, as I understand it. Adjusting it has an effect, immediately visible on the raw graph. Just do some tries and see.
As above said, keep the black & white clipping values at 0% or so.

About DNG, I have seen a massive improvement from Raw DNG over TIFF, when negatives are extreme. No idea why, the settings are identical (I export each scan simultaneously in jpeg+tiff+raw dng).
Here is an example of a very difficult scan, which was satisfactory only using Raw DNG (+ multiexposure + 8x multisampling).
(photo copyrighted - pls no reuse)