SilverFast Clarification Needed: Impact of Custom Settings on Auto IT8 Calibration Accuracy

[Archivist_Goals]

Hey all, I've been using FilmScanner.info for a long time. The site has been a goldmine of information for scanning. I have one question that still persists and perhaps one of you can help.

I am using the following for only scanning of reflective materials:

Hardware: Epson v850 Flatbed Scanner
OS: Windows 11 64-bit
Software: SilverFast Ai Studio 9 Archive Suite + (HDR Studio)
IT8 Target: LaserSoft's 'Advanced' Reflective Color Target that follows ISO 12641-2 (Part 2)

Scanning at:
48-Bit HDR RAW
2400 DPI
TIFF (RAW/Linear Output file)

SilverFast support has been slow to help, and can sometimes be quite vague and not helpful to explain the 'why?' behind more technical questions.

Regarding their Auto IT8 Target Calibration, they have informed me:

SilverFast Support:

"During calibration, SilverFast will automatically set the software to neutral values for the process.
You do not need to care about settings in the software for it."

Me: I understand that scanner profiling should match the settings used for subsequent scanning tasks. However, I've noticed that SilverFast resets to default settings (e.g., from my set values of 48-bit HDR RAW to 48 -> 24-bit, and from 2400DPI to 300DPI) when initiating the Auto IT8 calibration.

My remaining questions are:

1. If I revert the settings (Color Depth and DPI) to my preferred values (48-bit HDR RAW and 2400DPI) before the calibration begins - but after I click the Auto Calibrate button - will this affect the accuracy of the generated ICC profile?

2. Is the adjustment to 24-bit and 300DPI a compulsory step for the Auto IT8 calibration, or can I maintain my desired settings (48-bit HDR RAW and 2400DPI) throughout the calibration process without compromising its effectiveness for generating a proper and accurate ICC profile?

3. Also, I understand that HDR RAW files will be very dark. However, when opening in Photoshop, the histogram shows the shadows pushed to the extreme left, a lot of clipping.

Does anyone here know the answer to this?


PS. Once I create a proper ICC profile from the Auto IT8 Calibration, I will begin creating RAW scans of archival materials.

Workflow: SilverFast > Scan at 48-Bit HDR RAW, 2400 DPI, TIFF > Save > Open TIFF in Photoshop, apply ICC Profile > Save > Open COPY of that same file in Photoshop > Secondary Processing for Derivative Files for color adjustments, leveling, downsampling to 24-bit, etc.

But I am urgently requesting clarification on impact of custom settings during Auto IT8 Calibration, thanks!

[Jossie]

Good morning,

For prints a sampling rate of 2400 ppi is too high. Normally 600 ppi should be sufficient. You should test this by a series of scans with sampling rates between 600 ppi and 2400 ppi and the select the setting which just gives the optimum resolution.

Next, if LSI says your settings for the auto-IT8-calibration run are irrelevant that must be correct. You need not worry about these. SF is using its internal settings for the calibration run. This is how I understand their answer.

However, your workflow is flawed, if you open the HDR-scans in PhotoShop and use the ICC-profile created by SF. This profile is meant to be used for gamma-corrected images, as e.g. in SF HDR. This is the reason why the images in PhotoShop are very dark and the histogram is shifted to the left.

Finally, I recommend using Argyll to create the ICC-profiles. It gives you much better colours than the SF profiles.

If you intend to use PhotoShop instead of SF HDR anyhow, I do not understand why you have bought the expensive SF archive suite. According to the version comparison on the SF web pages you can save 48/16 bit HDR scans in all SF versions.

Hermann-Josef

[Archivist_Goals]

Hermann-Josef,

Thank you for your reply. I think this *finally* explains what I was doing wrong - and why I kept receiving super dark results, even after applying the ICC profile I created in SilverFast. The following is a long post. And I suspect you are based in Germany - I am in the North East, USA. So, please take your time to read and respond accordingly.

So, let me backtrack. Yes, I am scanning only reflective material. No transparencies.

I'm scanning:

Yes - Photographic prints (mostly color, some B&W)
--> Some as large as 8x10 (essentially maximizing the full scan area of the v850)
- CD-ROM & DVD optical discs (both label and data side; enough resolution to make legible mastering ring codes which are printed very small)
- Inserts, manuals, and other paper ephemera documentation
- Box Art
--> Historical and/or obscure software box art, cases, sleeves, etc.
- Original Newspaper Articles
- Vinyl single '45s' (7-inch) records

According to resolution tests done for the v850, the effective resolution for the unit maxes out around 2600DPI when using SilverFast Ai Studio. In this context, I had planned to safely scan everything at either 800DPI (for paper material) or 2400DPI (for optical discs and box art).

Yes, it is much too high for most average prints, and indeed, most average reflective material. ISO and most other cultural institutions (US Library of Congress, NARA, etc) have indicated 600DPI for most smaller format prints and paper materials. My goal it to exceed this standard as an attempt at 'future proofing' the scans. In short, yes, not everything will be sampled at 2400. Some will be sampled at 800 or 1200 depending on what they are. Newspapers, for example, won't benefit much from beyond certain sampling rates, as I am sure you know.

I am also scanning at 48-bit (or what LSI designates as 48-bit color HDR RAW) in which some people might find to be overkill or pointless. But, again, this is my attempt at future-proofing the scans to present the most detail possible with the hardware and software I have in a non-destructive way.

"Next, if LSI says your settings for the auto-IT8-calibration run are irrelevant that must be correct. You need not worry about these. SF is using its internal settings for the calibration run. This is how I understand their answer."

Thank you. But I am not sure if this answers my question. Let's say I have the following presets in SilverFast set:

My Preset Values:
Reflective
Positive
48 Bit HDR RAW
2400 ppi
No input profile

I click IT8 Auto Calibrate button and then SilverFast software automatically reverts my settings to the following:

SilverFast Preset Values:
Reflective
Positive
48 -> 24 bit
300 ppi
No input profile

Once SilverFast reverts to these new settings, it scans the target and waits for my input to confirm the process to begin calibration. However, I have the option to REVERT the settings it designated, and change them back to my settings - the ones I designated in 'My preset values' above.

Put bluntly, is this right / correct that I change it back to my settings? Or, is the right way to profile in SilverFast using their Advanced Targets, to accept what SilverFast changes the values to?

From what I understand, one is supposed to profile a scanner according to the values one will be scanning actual material at afterwards. From what I understand, if I am going to be scanning at 'My preset values', shouldn't I profile the scanner at those same values?

LSI says it does not matter. But then why does SilverFast revert the settings to much smaller values, e.g. '24-bit, 300DPI' for the calibration process?

Which is the right option?

I do not know if LSI support understood my question entirely. Again, they are always vague in their responses and do not explain the 'why', 'how', or elaborate in much detail. If my description of what SilverFast does during this process is not clear, I can upload a video recording and post it here. Let me know, and I'll be happy to do so.

"However, your workflow is flawed, if you open the HDR-scans in PhotoShop and use the ICC-profile created by SF. This profile is meant to be used for gamma-corrected images, as e.g. in SF HDR. This is the reason why the images in PhotoShop are very dark and the histogram is shifted to the left."

Thank you, this makes much more sense now that you pointed it out. To answer your question: Unfortunately, I purchased the Archive Suite without realizing the workflow that it implies, e.g. to use HDR Studio, too.

Workflow: So, I would scan to 48-bit HDR RAW output in SilverFast, and then open that file in SF HDR Studio > Apply 2.2 Gamma value > Save?

What should the workflow look like?

I do not have any experience using Argyll and would need help getting started with it. Perhaps you can assist? I have tried using CoCa, but it will not accept LSI's Advanced Targets for some reason.

As an aside, I spoke with Wolf Faust recently through email (I also have some reflective targets from him, the older ones for ISO 12641-1) and he informed me that the newer standard (LSI's 'Advanced Targets') have colors which, in his words, are "impossible to produce". He claims it is a not a good standard. And while I trust what he is saying, this raises an issue:

If I want to use the Advanced Target from LSI, CoCa doesn't recognize and presents with the following error,

"RGB values cannot be extracted. And to check the target type selection is correct."

I do not know if this is because CoCa can not read CxF data files (LSI only provides CxF formats for their Advanced Targets, which is dumb in my opinion), or for some reason it simply will not recognize the Advanced Target layout, even though it offers the option to select the target type for, 'LaserSoft Advanced 864 Patch'. I spoke with Dave Hamrick (Ed Hamrick's son) of VueScan software recently. And he informed me, that should I want to switch to VueScan instead, VueScan does not accept .CxF data file format.

I have *also* spoke with BabelColor for use of their PatchTool to convert the CxF to a TXT in hopes of importing into CoCa. But it still did not work and presented with the error I mentioned above. Ideally, I want to use the LSI Advanced Target for a wider color gamut. But, I am not sure if I am chasing such minute detail that it doesn't matter in this case... when comparing ISO 12641-1 vs ISO 12641-2.

Please let me know your thoughts and many thanks!

King Regards,
Nick (Archivist_Goals)

Edit: I am attaching a video recording of SF HDR Studio. When you open an HDR RAW file from SilverFast Ai Studio in SF HDR Studio, from what I understand you are saying that: One is supposed to set the "expected gamma value from the HDR/HDRi files" (which is 1.0 since SF produced it as a linear TIFF) and then set the actual Gamma Gradation to 2.20 and then - depending on if I apply the ICC in HDR Studio or not - 'Process the file' > Save?

But then can I take that saved file from HDR Studio and further edit/adjust Levels, etc., in Photoshop and maintain the adjusted 2.2 gamma value?

In other words, can my workflow be SF Studio Ai > SF HDR Studio > PS for further post-processing?

https://www.youtube.com/watch?v=CnhwhHJ1Pkg

[Jossie]

Hi Nick,

just a few points.

If you want to stick to SF IT8 calibration, just leave all the settings at their defaults (reset the software to be safe!) and run the auto-calibration. Then apply the profile to your scans with your favourite settings (scan rate will not really affect the calibration) and export the scan in sRGB without any tools applied. These exported images can then be further processed with PhotoShop.

I suspect that CoCa does not support the extended IT8-format, but Argyll does.

The other option would be to scan the target and export it in the HDR-format. Use these scans in Argyll to create an ICC-profile (be sure the scan size in pixels is not too large, otherwise Argyll will not work, 2000 x 2000 pixels will probably be okay) and embed this in your scans. This way the gamma-correction is included in the profile and you will have no problem in PhotoShop.

I hope this helps.

Hermann-Josef

[Archivist_Goals]

Hermann-Josef,

Thank you.

You say:

"Leave at defaults and/or reset the software and then run the auto-calibration."

I'm to understand that you don't know for certain, either? Seems like nobody knows the answer to this, and LSI doesn't want to elaborate.

You say:

"Export the scan in sRGB."

I've seen various people say, sRGB, some say Adobe RGB (1998), others say Wide Gamut RGB, etc. Does this come down to personal preference, or is sRGB superior to Adobe RGB in this context?

I recall seeing a front-end GUI for Argyll. Maybe I can give that a try, as Argyll command-line interface is quite daunting. But that does make sense regarding CoCa and not being able to support the extended IT8-format. What's confusing is the option is present in the GUI, like I mentioned, for 'LaserSoft Advanced 864 Patch' - Perhaps it was never properly configured, however. I recall development for CoCa stopped a number of years ago. And even thought about reaching out to the developer, Andrew Stawowczyk Long. But I can not find an easy way to contact him.

I will give Argyll a try.

Also, I'm assuming the workflow in the order of SF Ai Studio 9 > SF HDR Studio > Photoshop is not possible/practical (I think?) since one can only have the corrected Gamma value from HDR Studio show up in HDR Studio...and not export properly to Photoshop?

Which is to say, LSI has made it more difficult to have the aforementioned workflow. Rather, they try to impose their additional workflow, e.g. force one to use HDR Studio for processing. Hence, 'Archive Suite'. The problem with this is HDR Studio is not a professional photo editing software such as PS or GIMP. All I have wanted to do is create unaltered, uncompressed RAW data file outputs, apply a proper ICC from a good color profiling session and save those as Archive Masters. And then create copes of those Archive Masters and edit accordingly as so-called derivative files for aesthetic purposes.

If you have any further thoughts, they are welcome.

Many thanks for your replies and input.

Kind Regards,
Nick

[Jossie]

Hello Nick,

I'm to understand that you don't know for certain, either?

Well, I know it is not critical which sampling rate you choose as long as it is within a reasonable range, and if LSI has decided to go with only 8bit that may explain in part why their profiles are not as good as Argyl's.

Does this come down to personal preference, or is sRGB superior to Adobe RGB in this context?

No, it depends on what you plan to do with the scans. sRGB is the smallest colour space, AdobeRGB is larger, ProPhoto the largest. For web applications you should use sRGB. I recommend that you archive and keep your HDR-scans so you can always go back to them. The colour space of the scanner is usually quite large.
My main concern in this respect is the question of the colour space of your prints. I have no idea how large a colour space they require. I would assume, that the prints' colour space is much smaller than that of slides / colour negatives. So this leads to the question which effort is worth for your purposes...

I'm assuming the workflow in the order of SF Ai Studio 9 > SF HDR Studio > Photoshop is not possible/practical

This is not what I said nor what I meant. If you want to use PhotoShop for all purposes you need to create an ICC-profile with Argyll, which has the gamma-correction incorporated. Then you do not need SF HDR at all since you can export the HDR-Scan from Ai with this profile embedded. And there you go with PhotoShop. Since you are only concerned with scans of prints (and not transparencies) you need not worry about iSRD, the IR-based dust and scratch removal. This is only possible with SF HDR.

All I have wanted to do is create unaltered, uncompressed RAW data file outputs,

See above...

Best wishes

Hermann-Josef

[Bobby]

Why do you use the "Auto IT8" feature? I always used the manual IT8 calibration (Tools/Scanner/Standard Calibration), which is very easy. I do not trust such automatic features *, as you have no control on what is really happening, whist with manual calibration you actually see what is going on. I now use Argyll profiles. They are more challenging to create, but give definitely better pictures. But the Silverfast IT8 profiles are also oK if your requirements are not too high.

* example is the "Automatic" feature in iSRD. If you use this with Kodachrome slides, your picture content details will be severely disfigured. Here, manual processing is required.

By the way, I have problems understanding why you want to make document scans with such high resolution. Do you want to blow up stamps to poster size? What resolution do your documents have, and how sharp is the focus of your pictures? Normal photo prints rarely deserve better scanning than 600 dpi. Your scans do not need better than the original.