Use several exposures in scanning film to increase information?

[Erland]

Hello!

I have started to scan my film, black and white, most often Tri-x. I want to increase the information in my files, using several exposures in Vuescan. Mostly to get better gradients, like increased apparent bit depth? More so than increase the dynamic range. Is this possible?
I have searched the forum and googled, but google seems to prioritize newer threads rather than old about scanning.

I scan in 48bit raw tiffs, using a Plustek with vuescan. My histogram has a lot of information, but the tiff lacks any ability to adjust exposure and just using a simple levels adjustment's middle pointer, the image gets crushed rather quickly.

Kind Regards
Marcus

[Mark D Segal]

Hello!

I have started to scan my film, black and white, most often Tri-x. I want to increase the information in my files, using several exposures in Vuescan. Mostly to get better gradients, like increased apparent bit depth? More so than increase the dynamic range. Is this possible?
I have searched the forum and googled, but google seems to prioritize newer threads rather than old about scanning.

I scan in 48bit raw tiffs, using a Plustek with vuescan. My histogram has a lot of information, but the tiff lacks any ability to adjust exposure and just using a simple levels adjustment's middle pointer, the image gets crushed rather quickly.

Kind Regards
Marcus

If you are scanning in 48 bit mode you are getting 16 bits per channel which is all the bit depth available (provided Vuescan is not letting you make settings the scanner can't actually produce (which it shouldn't) - you didn't say what scanner you are using). I presume these are negatives you are scanning. The negatives need to be inverted to positives. Make sure the inversion process isn't clipping image information - look at the post-scan histogram of the positive rendition. In principle, using good software you should be able to make very fine luminance adjustments for both contrast and brightness. Have you tried processing those files in Lightroom or Photoshop?

[Bart_van_der_Wolf]

Hello!

I have started to scan my film, black and white, most often Tri-x. I want to increase the information in my files, using several exposures in Vuescan. Mostly to get better gradients, like increased apparent bit depth? More so than increase the dynamic range. Is this possible?

Hi Marcus,

First of all, negatives like TRI-X usually do not have a high enough optical density to create Dynamic range issues for most scanners. So you do not need to resort to multiple exposures per scan line, or multiple image scans, for increasing the dynamic range of the scan. In fact, you run a risk of mis-alignment due to film expansion during the scan, caused by an increase in temperature. That would create a risk of mis-registration of the subexposures.

I scan in 48bit raw tiffs, using a Plustek with vuescan. My histogram has a lot of information, but the tiff lacks any ability to adjust exposure and just using a simple levels adjustment's middle pointer, the image gets crushed rather quickly.

VueScan will produce either 16-bit per channel grayscale or 48-bit RGB image files, and they are already inverted for normal output viewing. It can even produce 64bit RGBI data files from scanners that have an IR-light dust detection feature, but that won't work with silver based film, because silver is opaque to IR light. Only Chromogenic B/W film could benefit from that. The benefit of scanning B/W film as RGB is that you can later average the three channels which will reduce noise. VueScan has some controls (Black/White-point, s-curve, and a brightness control) for postprocessing of that data, but it's obviously only a small sub-set of what e.g. Photoshop can do with the output file.

Cheers,
Bart

[Erland]

I have a plustek 7300. It is a basic model, without ICE. Also from what I've read it has a pretty lousy dmax and Dmin.
Vuescan is set to produce a raw file, in tiff format instead of dng. It is an untagged tiff, 16bit per channel, but it does clip either the blacks or the whites depending on the exposure used in Vuescan (possibly because the low dmin?). I then invert it in photoshop using a simple invert command, and then desaturate it with an HSL layer.

I have read that even though it produces a 16bit file, the scanner might be an actual 8bit, interpolating to 16bit and I actually think it might be so. Hence my question about scanning multiple times, using different exposure perhaps, and combining them in photoshop to get better gradations.

Why I really ask is because when I do an actual darkroom print of the negative, I get more tones on Ilford MG paper, that I can't reproduce when scanning.

I have attached a file from yesterdays session. It is a jpeg made from the final tiff.

[D Fosse]

I shot a lot of Tri-X back then, it was my absolute favorite because of the extremely crisp grain.

So with the disclaimer that I haven't for a long time tried scanning Tri-X negatives: My immediate thought is that if the scan is really good, you'll get a "binary" image without true gray values, only black or white. But the histogram or previews won't necessarily reflect this because it works on cached data. This is a classic question in the Adobe Photoshop forum (why do I lose adjustments <in my noisy/grainy image> when flattening?)

Another implication of this is that scanning resolution plays a role, perhaps more than bit depth. The negative is basically 1-bit, on or off.

[Mark D Segal]

I have a plustek 7300. It is a basic model, without ICE. Also from what I've read it has a pretty lousy dmax and Dmin.
Vuescan is set to produce a raw file, in tiff format instead of dng. It is an untagged tiff, 16bit per channel, but it does clip either the blacks or the whites depending on the exposure used in Vuescan (possibly because the low dmin?). I then invert it in photoshop using a simple invert command, and then desaturate it with an HSL layer.

I have read that even though it produces a 16bit file, the scanner might be an actual 8bit, interpolating to 16bit and I actually think it might be so. Hence my question about scanning multiple times, using different exposure perhaps, and combining them in photoshop to get better gradations.

I have attached a file from yesterdays session. It is a jpeg made from the final tiff.

To start with, ICE will not work with B&W film, so no loss there.

If the scanner can't deliver 16-bit per channel output native, you are depending on interpolation to invent information, and that is second-best. If the scanner program is causing information to be clipped, you can try using different exposures and combining the results in Photoshop using masking and merging. A lot of work, and I think Bart's cautionary note on this is worth considering. Check all your Vuescan settings and make sure it is actually rendering a "raw" scan (i.e. capturing all the tonal information the scanner can deliver). And why are you not letting Vuescan invert the negative? You may be causing problems for yourself doing this in Photoshop. (I have Vuscan, which I don't use often, but I see no reason why it shouldn't be able to render whatever your scanner can deliver if the settings are correct.) I do use SilverFast as my preferred application and I have no doubt that it will render all the information the scanner can deliver and produce a suitable inversion for further work. You can download a free trial version of SilverFast provided there is a version for your scanner. This will help you see whether the problem you are having is scanner-constrained or application-constrained. But first review your Vuescan settings to make sure they aren't compromising information and do the inversion in Vuescan. Inverting in Photoshop can be made to work well - I've experimented with numerous approaches for doing this, but the first option should be to do it with the scanner software.

[Bart_van_der_Wolf]

I have a plustek 7300. It is a basic model, without ICE. Also from what I've read it has a pretty lousy dmax and Dmin.
Vuescan is set to produce a raw file, in tiff format instead of dng. It is an untagged tiff, 16bit per channel, but it does clip either the blacks or the whites depending on the exposure used in Vuescan (possibly because the low dmin?). I then invert it in photoshop using a simple invert command, and then desaturate it with an HSL layer.

Maybe it's better to not produce a Raw file (linear gamma inverted like the film), but let VueScan invert the image, and apply a gamma curve. But first of all you need to let Vuescan expose the film correctly. There should not be (any need for) clipping of either shadows or highlights, but you then must set the Black and White point percentages on the Color tab to zero, or very low values, to avoid post scan clipping of the full set of data.

Cheers,
Bart

[Erland]

Mark, I bought the scanner second hand, but it Came with silverfast. Just it said something about it was already registered?
Will try it again perhaps.

The reason vuescan isn't inverting the image is since it is raw file, it doesn't do anything to the tiff, neither mirroring nor inverting.
But I will try letting Vuescan create a readymade file next time.  I can select it saving that as a tiff file as well, so I don't have to worry about compression or lowering the bit depth. However I can't see that this would make my files contain more tones? Yes, I do believe the scanner is the culprit here, and instead of investing in a better one (amateur photographer shooting film for fun) I'd rather see if I could make it better using simple tweaks.

[Mark D Segal]

Mark, I bought the scanner second hand, but it Came with silverfast. Just it said something about it was already registered?
Will try it again perhaps.

The reason vuescan isn't inverting the image is since it is raw file, it doesn't do anything to the tiff, neither mirroring nor inverting.
But I will try letting Vuescan create a readymade file next time.  I can select it saving that as a tiff file as well, so I don't have to worry about compression or lowering the bit depth. However I can't see that this would make my files contain more tones? Yes, I do believe the scanner is the culprit here, and instead of investing in a better one (amateur photographer shooting film for fun) I'd rather see if I could make it better using simple tweaks.

I would try not using it "raw" then, but do as Bart suggests - making sure the settings aren't clipping highlight or shadow information, and let the scanner software do the inversion. See how that helps. As long as the software settings aren't clipping highlight or shadow information you will get all the information the scanner can deliver if set to 48 bit mode. If highlights or shadows are being clipped, by definition you are compromising on information. If you scan at lower bit depth you are compromising on the extent of luminance information across the board.