Silverfast questions

[mani]

Once again I'm giving Silverfast a try - I downloaded the trial a while ago but didn't manage to get it activated (probably because I'd used the trial a couple years before on the same computer), but now it worked, so I have 30 days to give it a thorough investigation. There will be a few questions.

For these tests I'm scanning on the Nikon Coolscan LS9000.
I am not using any glass holders - so no artefacts are Newton rings.
I am scanning at the scanner's optical resolution (4000ppi)
I am scanning 48bitHDR Raw linear negative files with a gamma gradation of 2.2 to match my settings in ColorPerfect
I am using Multiple Exposure - which is a second pass for extra dynamic range
All other image manipulation tools are switched off when using the 48bitHDR Raw setting

The first question:

I've noticed that if I have an area that's extremely over-exposed in my negative (I'm often pushing my exposures to +4 or +5) then that area of the scan has a VERY subtle 'fingerprint whorl' wave pattern (see image of a tiny area of a negative zoomed to 600%). The line I drew on the image didn't really follow the wave so well I see now - sorry.

At normal viewing resolution this isn't really visible, but it becomes visible when zoomed to 200% and upwards.

Now I honestly don't need to be told that this is irrelevant to the finished print or to 'stop pixel-peeping' etc. I'm trying to make a decision whether to buy the software (which costs around €400 for the LS9000 I believe) and I'm simply trying to investigate possible problems. So if anyone has seen the same artefacts in their scans and know the reason/cure, then I'd love to hear about it.


PS: I've written to Silverfast, but haven't received an answer yet. But I'm more interested in other users' experiences, in any case.

 

[Larry Cloetta]

That's not a problem I have ever had with Silverfast, though my needs might be different, in that if I turn up with a negative that is 5 stops off I usually bin it without bothering to scan it. On the rareish occasions I might be scanning something which had areas that overexposed, I don't recall similar artifacts occurring. FWIW.

 

[mani]

Thanks Larry for the input!

I wouldn't normally bother with this particular negative, but I noticed the artifacts also appearing in (say) a highlight on a persons bare arm in very bright sunlight in a normally exposed photograph, so it's not only restricted to grossly over-exposed negs - just more obvious when the bright area is over a larger section of the image.

Incidentally, the digital color noise is really easy to clean-up, but the luminance wave pattern can't be removed without damaging detail.

I'm being a bit of a perfectionist here - hurts me to sink those 400 euros on this particular software. 🙁
PS: don't mean to imply others aren't being perfectionists! Just meant that I'm actively looking for problems before buying SF!

 

[brbo]

...then that area of the scan has a VERY subtle 'fingerprint whorl' wave pattern (see image of a tiny area of a negative zoomed to 600%)

I wouldn't be judging ANYTHING at 600%. Who knows what goes on in the resizing algorithm... View at 1:1 (100%)!

Make sure you have ALL data altering settings turned off. Turn off multi exposure (only "benefit" you get with negative film is another aligning and pixel binning algorithm that does whoknowswhat), 2.2 gamma for raw (ColorPerfect is perfectly capable of working with linear files) and 'limit gamma slope'.

edit: that is, of course, if Vuescan doesn't exhibit the same problem...

 

[mani]

Received a very prompt reply from Silverfast - maybe their customer care is improving!

"you can try to switch to the second scan mode your Nikon scanner provides:
Perform three line scan with optical resolution inside the SilverFast 8
Preferences -> Special
If this does not help, please check if there are electrical devices around your
scanner that might cause electrical disturbances (fluorescent tubes, for example).
We have also some customers which report, that deleting the FireWire
Network connection (Mac only) improves the quality of the scans."

The pattern is definitely not a display scaling artifact - not only is it clearly visible in different applications and different scale ratios, I've also scanned the images at exactly the same size and values with Vuescan, so I make 1:1 comparisons.

Anyway, I knew I'd get some flack for looking at the images at more than 100%, but I definitely don't wanna spend the next few years scanning my negatives, only to find that I can't make big prints of the resulting negatives because they possibly exhibit some weird fingerprint wave pattern in bright patches.

 

[Ronald M]

In the past, I have scanned problem negatives with multiple exposure. Problems include overexposed exteriors on interior shots.

Usually 2 scans will do it, one for the basic exposure, one to get a correct exposure for window or bright area.

Treat as a HDR and you do not have to jack up contrast/saturation so it looks like HDR. People can not tell mine from normal photo. Treat the neg as 3 or 2 separate digital files.

Arrow keys will align stack one pixel at a time or try automatic feature. Reduce opacity of top layer so you can see what you are doing.

 

[mani]

I scanned six more images tonight, testing with different settings, and found that the "Perform three line scan with optical resolution" actually helped with the subtle wave-pattern artifacts - though maybe it lost some highlight color nuance.

I have to say the most impressive thing about the Silverfast scans compared to the Vuescan scans, is that they look the way I want straight out of ColorPerfect - without any of the extra tweaking I always need to do with the Vuescan images (whilst both of them are supposedly flat linear 48bit negative scans).

Oh - and the fact that the SF scans (even when using multi-passes) take a fraction of the time of Vuescan.

 

[brbo]

I have to say the most impressive thing about the Silverfast scans compared to the Vuescan scans, is that they look the way I want straight out of ColorPerfect - without any of the extra tweaking I always need to do with the Vuescan images (whilst both of them are supposedly flat linear 48bit negative scans).

That is because you are tricking the ColorPerfect into thinking that your 2.2 gamma encoded "raw" scan (it's clearly not unmodified-straight-out-of-scanner raw anymore) is linear. And I agree, I often find them better looking (ColorPerfect color integrity be damned 😉). But if you do it properly, linear raw and 2.2g "raw" will come out exactly the same.

Btw, you can produce gamma encoded "raw" file just as easily in Vuescan. You can even do that in PS before you feed the raw scan to ColorPerfect if you like ColorFerfect output better this way.

Oh - and the fact that the SF scans (even when using multi-passes) take a fraction of the time of Vuescan.

That's odd, I haven't noticed that with my CS8000 when I had it. Vuescan, with exposure locked at 1.0, should be just as fast as Silverfast. But if you don't lock exposure, Vuescan will increase exposure time to maximum length that still doesn't clip the low density areas on the film (and with very dense negatives that can be very very long).

It is possible, though. I've seen something like that with another scanner (Vuescan and Silverfast scanning regimes very different to manufacturer's scanning software).