Silverfast 8 giving inferior shadow detail results to v6.6

[Anorakus]

Hi,

I've just upgraded to Silverfast 8 from v6.6. In both cases I'm using 64-bit Windows 7 and Adobe Photoshop CS5.

I've found that Silverfast 8 is resolving far less shadow detail that v6.6.

Here are two examples. The first is scanned with v6.6, the second with v8. The original is a slide taken in the 60's. There is a magenta colour cast, however viewing the slide by eye shows plenty of detail in the shadow areas under the table.





The area under the table shows detail in the first scan. In the second, the detail is lost and I'm left with a green cast.

In both cases, automatic colour correction and multi-exposure were applied in Silverfast (Auto in v6.6, Auto CCR in v8). GANE and auto-contrast were also applied in v8.

I've played around with various combinations of settings, but I am unable to resolve shadow detail that I was achieving in v6.6.

I am about to start a big restoration project, and I am ready to downgrade back to v6.6, as from what I have seen so far, v8 is producing inferior results.

I am using a Plustek 7600i SE Plus scanner, running on Windows 7 64-bit.

Any ideas welcome

Thanks,

-- John

[LSI_Morales]

Hi John

Are you working with the 32 or the 64 bit version of SilverFast 8?

Thanks

[Anorakus]

I'm running the 64-bit version of Silverfast, although I've since tried installing the 32-bit with the correct drivers, and it gives the same results.

I've since tried doing scans using the Plustek scanner by itself, i.e. using the QuickScan button. Here is an example - this is straight from the scanner, with default colour correction selected in the scanner driver settings. As you can see, shadow detail is much better than the Silverfast 8 scan, even without multi-exposure:



I note that the colours are more realistic too - the Silverfast scans show a strong red tint to skin tones in shadow, particularly noticeable on the lady's legs under the table. This is absent from the scan done straight from the Plustek.

-- John

[LSI_Morales]

Hi John,

thanks for your follow-up we are carrying out test and are investigating this issue. As soon as we have more information on any progress we will post it here.

Kind regards

[DavidP]

I have the same hardware and software environment as Anorakus. I am experiencing a very similar problem with detail in dark areas of slides (especially, of course, with Kodachrome).

A fix would be very much appreciated!

Thanks,
David

[LSI_Ketelhohn]

Dear David,

Thank you for your input.
I will inform my colleagues.

kind regards,
Arne ketelhohn.

[theprincereturns]

Nothing more to add other than that I am expiriencing this with my Plustek 7600i as well.

[snäckan]

I have also problem to get the correct colors and details in silverfast 8.
Silverfast 6.6 is better to get the correct colors.
I have also Plustek 7600i

[Anorakus]

Hello,

I have installed the latest r5 update for 64 bit Silverfast with my Plustek 7600i. Unfortunately it seems to have made no difference to the results I am getting. Here are some sample scans of an old slide that has faded to a strong magenta colour cast, typical of the restoration project I am working on:

#1 is a scan at 3600 dpi using Silverfast. Auto CCR, auto contast, GANE and ME are turned on:



#2 is a scan done directly from the Plustek driver using the QuickScan option, i.e. Silverfast not used at all. 3600 dpi and auto colour selected:



#3 is scan #2 with Photoshop auto levels (including snap to midtones) applied:




As you can see, scans #2 and #3 resolve far more detail in the tree branches and dark areas than #1. Scan #1 also has a green tint to the black areas which is difficult to remove without altering the color balance of the whole scan.

Additionally, the r5 release took about 10 minutes to scan one slide with the above settings applied!

I cannot use Silverfast at all in its current state - it produces very poor and unacceptable results. Plustek's own scanner driver produces much better scans.

Can you reproduce these results in the lab and fix the software? It is very frustrating to have paid €60 for a piece of software that doesn't work properly.

Feel free to contact me directly if you wish.

-- John

[LSI_Morales]

Hi there

Thanks for posting

As you can see, scans #2 and #3 resolve far more detail in the tree branches and dark areas than #1. Scan #1 also has a green tint to the black areas which is difficult to remove without altering the color balance of the whole scan.

This issue is far more complicated than it appears.

The images which all you have been providing as "more detailed" are images created without any color management. If you activate the color management option in QuickScan you will obtain similar (or even inferior) results than those produced in SilverFast.

You can also turn off the SilverFast color management System in Ai Studio and will see how the outcomes are pretty close (or even "better") to that of QuickScan. The great problem is, the resulting pictures are not color managed.
We are looking into possible ways to improve these issues without having to sacrifice the whole color management system, but these are not only software related problems.

Can you reproduce these results in the lab and fix the software? It is very frustrating to have paid €60 for a piece of software that doesn't work properly.

Yes we have seen this behavior in the lab but as I mentioned before, the software is working properly, there is a workaround for this by turning off the color management but then you will have to make a sacrifice.

Improvements on this might take a while since we are experimenting with many different approaches, but unfortunately have not had any way to obtain the absolute quality without compromising any of the two variables.

Any contributions and ideas on this respect are very welcome.

Thanks you

Kind regards

[Anorakus]

Thanks for the reply

What I really want is for Silverfast to resolve all the available information in the slide - in other words, I need the best dynamic range possible. Colour management isn't important to me at the scan stage - I can do this in Photoshop, which is a far more powerful tool for adjusting colour than any scanner software could be.

Obviously if dynamic range is lost at the scan stage, there is no way of recovering it in Photoshop. This is why the scans using QuickScan appear better than those in Silverfast - they are resolving shadow detail which is lost in the Silverfast scans.

Perhaps colour management in Silverfast isn't suited to old slides with strong colour casts. I will try again with colour management turned off and let you know how I get on..

-- John

[LSI_Morales]

Hi there

What I really want is for Silverfast to resolve all the available information in the slide - in other words, I need the best dynamic range possible. Colour management isn't important to me at the scan stage

Actually if that is what you want, you should scan in 48 bit HDR format with Multi-Exposure on, which will get much more details from the film in a raw file.
Since this is a raw file, you can open it in photoshop make gamma corrections and assign any color profile you want as well as any other corrections.

Kind regards

[Anorakus]

OK, I've tried the HDR option.. once I found out to tick the "HDR output" box on the gamma settings option screen, I am now getting much better results Here are the two slides above - in each case I've used Photoshop auto-levels to correct colour, then adjusted midtones and contrast to give a pleasing result:





I think perhaps it should be stressed more that HDR scanning and then Photoshop or HDR studio is the solution for faded or underexposed slides. The auto options only seem to work well for good quality source material.

But thanks anyway - I can get to work on my project now!

A.

[LSI_Morales]

Hi Anorakus

I am glad thins are now working better for you and the proposed solution is also valid for your workflow. See how Multi-Exposure and the HDR raw files improved on your pictures more than what you would get from other applications?

Kind regards