2nd Thoughts on Silverfast Scan Software

Cool! Hope you're outside in the warm evening! (Just got back from Hagaparken with the kids).

Which Plustek scanner are you using?
Yes very much. Did similar, out barbecuing with the kids and friends.

I have the 8200i SE which has the more basic version of SilverFast I believe. It's the SilverFast 8 I have. Not really sure what it's lacking that the other versions have.
 
have the 8200i SE which has the more basic version of SilverFast I believe. It's the SilverFast 8 I have. Not really sure what it's lacking that the other versions have.

OK - I'm not certain myself, but annoyingly Lasersoft disable some of the features in the bundled Silverfast versions. I would suggest checking their website, but it's hard to tell even looking at the information there.

This blogger also has a few posts about Silverfast:
http://www.sebastian-schlueter.com/blog/2017/2/10/how-to-make-a-linear-scan-with-silverfast-88
 
So... I could make this a blind test, but since the files are so similar there is no point. Lets see who can come up with the best reason why the Silverfast files are the 'cleanest' (I can't and I looked hard).

Silverfast "raw" (it's not really raw as Silverfast manipulates the raw output + it's multi-exposure scan):



Vuescan "raw" (to basically match what Silverfast does to a raw scan, but it's not multi-exposure and no multi-sample, took about 1/3 of a time that Silverfast needed for previous scan):



Vuescan true raw (separate RGB channel exposures set so that the orange mask is removed at scan time (makes inverting without Colorperfect quite easy), no multi-exposure, no multi-sampling):




Inversions in Colorperfect (no post-processing after that except dust removal)...

SF:



VS (from "raw"):



VS (from true raw):




Do we have a clear favourite? Can you see the benefits of using multi-exposure with negative film?


As a bonus, I also did a scan in SF with NegaFix profile for Portra 160NC (the film is actually Vision3 500T) that gives quite a nice job with just a few tweaks:



My normal post processing (the raw scan from VS inverted in Colorperfect) using only Lightroom would look something like this:



Using only Photoshop (no Colorperfect) on VS true raw:




My conclusions:
- multi-exposure is not needed with negative film, could help with scanners that have more noise (but multi-sampling is better at that, too bad Silverfast got rid of multi-sampling in v8)
- Silverfast is not faster
- Silverfast does not produce better files
- if Silverfast profiles work (consistently) for you (they don't for me) and you like them they can save you quite a bit of time; I do like them from time to time - for this test I think the Portra 160NC profile did very very well so I tried another film and all profiles were totally off (reminded me why I gave up on Silverfast for quick hassle free scanning software that you could set on "autopilot"; and no, Vuescan can't be used like that either)
 
brbo, how did you get Vuescan to remove orange mask? I thought I knew the correct procedure, but could never get it to work.
 
brbo - thanks for taking the time and effort for your comparisons. Very interesting. I'm also curious about the orange mask removal in Vuescan - I'm assuming it's not just through the analog gain inputs?

In any case, it looks like your experience with Vuescan and Silverfast is different to mine.
As I said, I'm just using the settings suggested on the links I pointed to - no color profiles and no conscious image-manipulation from within the scanner software, with the exception of multi-sampling in SilverFast, which I've found helps with my (intentionally) over-exposed negatives.

In my case, the output from SilverFast takes a fraction of the time taken by Vuescan. The final image is easier to work with and the quality is better - in my own experience. Your experience appears to be different and I'm sure you can teach us a lot about the processes you use to get to the final image. I'm gonna leave it at that from my side though, as I haven't found that people discussing these matters with you in the past have really got beyond "you're doing it wrong".
 
brbo, how did you get Vuescan to remove orange mask? I thought I knew the correct procedure, but could never get it to work.

brbo - thanks for taking the time and effort for your comparisons. Very interesting. I'm also curious about the orange mask removal in Vuescan - I'm assuming it's not just through the analog gain inputs?

No, it's not just by using analog "gain" (it's not really gain, but individual RGB exposure). You can also remove orange mask on scanners that don't have individual RGB exposure capability.

In 'Input' tab select 'Media' = 'Image' or 'Slide', in 'Color' tab set 'Color balance' to 'None'. Make a selection of a part of the negative that is not exposed, Preview, check Lock exposure in Input tab (I usually lower the exposure that Vuescan suggests), Preview again. In the 'Image' menu set the 'Graph raw'.

If you have scanner that is capable of setting individual RGB exposure, change the individual R, G, B analog gain values and do previews until the graph shows that individual RBG channels overlap.



If your scanner can't set individual RBG exposures, right click somewhere inside your selection of clear part of the negative. This will set the orange mask as neutral grey. Look in the Color tab. Color balance is now set to 'Manual'. I'd suggest lowering the Brightness from 1 to 0.5 to give you some cushion in the shadows.



Or you could set the Media type to 'Slide', preview unexposed part of the negative, Lock exposure, Preview, Lock film base color, Lock image color. (Note that result will be different than what you get when you set grey point)



Remember that if your scanner can't set individual RGB exposures, your true raw file will still have the orange mask. You have to save to Tiff in your 'Output' tab. This is then your "raw" file with orange mask removed. Tiff is processed by Vuescan, so possibly every option you have enabled or set in 'Filter' and 'Color' tab will have effect on the result.

In my case, the output from SilverFast takes a fraction of the time taken by Vuescan. The final image is easier to work with and the quality is better - in my own experience. Your experience appears to be different and I'm sure you can teach us a lot about the processes you use to get to the final image. I'm gonna leave it at that from my side though, as I haven't found that people discussing these matters with you in the past have really got beyond "you're doing it wrong".

If it's of any consolation I'm often using Colorperfect the "wrong way", too (and I explained that to you in the other thread). And the "wrong" comes from Colorperfect guys. Since it's their software maybe they know what they are talking about?!

As for the lack of my contribution in scanning threads that don't go past "you're doing it wrong"... that's just ridiculous.

Now, would you consider doing a similar comparison of Silverfast raw (ME vs non-ME) and VS raw on your Coolscan? I'm curious what Silverfast can do with a Coolscan that it can't with my scanners.
 
Thanks for the detailed description for removing the orange mask - this is a much better description than the one I tried before, which was some blog post from a few years back. My problem was always that each image - even on the same roll - seemed to need tweaking again for the orange mask - but your way seems more robust. If I can't use ColorPerfect (for whatever reason) in the future, then I'll try this technique, because I'm never completely happy with other ways of removing the mask in post-processing.
 
Back
Top Bottom