Scanning and Levels

[Stealth3kpl]

Advertisement (gone after registration)

Is it better to set levels in the scanning software then scan and further modify in Photoshop, or is it ok to not adjust scanning levels, just output a file on which the black and white points haven't been set then set them in Photoshop? Does it make any difference? I've been doing the latter, but I'm just wondering if setting the black and white points prior to scanning mean more info is collected?

Pete

[Advertisement]

The Leica Q3 is available for pre-order at B&H Photo Video and Adorama

[Guest]

Hello Pete,

 

the scanner operates with a given bit depth, mostly 16bits/channel for good scanners. This is a fixed hardware property. So if you set the white and black point, either in the scanning software or later in PS, just resamples the data between white and black as set. So there is no difference in the result. In any case, the resampling will produce gaps in the histogram. Please note, that Sascha Steinhoff in his book "Scanning negatives and slides" states the contrary. But I think he is not correct.

 

However, if you want an optimum signal and your scanner hardware supports this, you should adjust the exposure time or lamp brightness of the scanner to use the full range / not overexpose the scan. I usually look at the histogram of a scan of an empty slide. The histogram should not severely saturate at the bright end but just reach towards the maximum value of 65535 for 16bits/channel.

 

Hermann-Josef

Edited October 2, 2018 by Jossie

[250swb]

I make sure the black and white points are set and then adjust levels to give a low contrast scan, it is ok for this to look horrible. Don't try to nail the scan with the scanner software, create something that gives all the tones then adjust in Photoshop/Lightroom after.

[Guest]

Best thing to do would be to adjust the "analog gain" (corresponding to exposure time or LED brightness, depending on the hardware) so the histogram is perfectly covered, if this is supported by the hardware. Then scan in what they call "raw"-format (althought it is not the type of raw we know from digital cameras). This demonstrates that the setting of white- and blackpoint in PS is perfectly okay. If analog gain is not available, just set the white point to the maximum (255 for 8bit/channel, 65535 for 16bit/channel) and black to 0 during the scan and do the rest in PS.

 

Hermann-Josef

[Stealth3kpl]

Thanks guys.

Pete

[philipus]

Hi Pete

 

The tricky thing in for example Vuescan is that while one can pull the edges of the histogram for each image to ensure that nothing is clipped or crushed, doing so doesn't work when batch scanning (of course). 

 

I only use the scanner to get as flat a scan as possible and then adjust during post-processing. This often means that the histogram looks a bit strange, either compressed or very wide or pushed to one or other side. But that can be fixed easily in post.

 

br

Philip

 

Is it better to set levels in the scanning software then scan and further modify in Photoshop, or is it ok to not adjust scanning levels, just output a file on which the black and white points haven't been set then set them in Photoshop? Does it make any difference? I've been doing the latter, but I'm just wondering if setting the black and white points prior to scanning mean more info is collected?

Pete

[Guest]

Advertisement (gone after registration)

But that can be fixed easily in post.

 

 

If we are talking about image quality this is not true. Expanding a narrow histogram in e.g. PhotoShop increases the contrast, but it does not improve the image dynamics. This means, that the number of grey levels cannot be increased by image manipulations. And the image dynamics determines the fidelity of the image. The only way out, as mentioned above, is to adjust lamp brightness or exposure time, if hardware permits. Setting the black and white point in the scan program does not resolve this issue since it also will not change the dynamics of the image.

 

Hermann-Josef

[Stealth3kpl]

 adjust lamp brightness or exposure time

 

I don't think this is possible with the Plustek 8100 and vuescan.

Pete

[Guest]

Hi Pete,

 

if so, I would say that setting black- and white point in the scan program is equivalent to setting it later in an external image editor.

 

Hermann-Josef

[250swb]

If we are talking about image quality this is not true. Expanding a narrow histogram in e.g. PhotoShop increases the contrast, but it does not improve the image dynamics. This means, that the number of grey levels cannot be increased by image manipulations. 

 

Of course you can't increase the 'number of grey levels', they are all within the black and white points, you can't invent more and can only work with what you have scanned. So you have 255 from black to white, and to avoid local clipping you start with a low contrast scan and make it higher contrast, you cannot do it the opposite way around by making a contrasty scan lower contrast because that is when you do get clipping. You don't lose anything you have scanned if you do it correctly and can get a perfect tonal range in the final post processed image. Nobody is suggesting you work with a 'narrow' bunched up histogram, you should have a long smooth histogram with no obvious spikes.

Edited October 4, 2018 by 250swb

[philipus]

It's evident one can't make something from nothing but ultimately it all comes down to what one is happy with in terms of the final image. Pretty much all image editing is destructive.

[frame-it]

i use the multi-exposure option on the plustek/silverfast...& occasionally the HDR option for b&w scans...its better to get as much information as possible into the image file while scanning...one cannot add it later

 

 

 

I don't think this is possible with the Plustek 8100 and vuescan.

Pete

[Guest]

@250swb

The question was, where it is best to set the white and black point. If the scanner does not permit adjustment of exposure time / LED brightness, there is nothing you can do to get a better dynamics than what the scanner delivers. Of course you can spread the existing grey levels between 0 and 255, but that only increases to contrast. What do you mean with "starting with a low contrast scan" if you cannot change the scanners exposure???? 

@frame-it  Do you think you gain something be using ME? The examples on the LSI Webseite are promising, but this is not at all what I get with my DigitDia6000. I gain nothing at all using ME. On the contrary, ME severely changes the brightness ratios in the image compared with the single scan, because neither SF nor VueScan remove the bias of the CCD, which is on the order of 650 for 16bit data (see section 4.1.5 in my remarks on SilverFast). If this also holds for the Plustek I cannot tell. Yes, if you are scanning into a HDR format you exactly see which dynamics the scanner delivers. The rest is image processing.

 

Hermann-Josef

[frame-it]

i only used ME 48>16 on some b&w acros 100 scans..the results were very nice.

[adan]

@250swb

The question was, where it is best to set the white and black point. If the scanner does not permit adjustment of exposure time / LED brightness, there is nothing you can do to get a better dynamics than what the scanner delivers. Of course you can spread the existing grey levels between 0 and 255, but that only increases to contrast. What do you mean with "starting with a low contrast scan" if you cannot change the scanners exposure????

 

Most scanners automatically adjust exposure time, to achieve whatever brightness you ask for in the controls. At least any decent scanner.

 

Sometimes you can even hear the difference, with an extra dense original: stepstepstepstepstep becomes step-step-step-step-step as the scanner leaves the scanner CCD bar under each "strip" of the film for a longer exposure time. With a very rapid scanner, the steps are not as distinguishable, but the total scan times for a dense or thin original can be compared - 30 secs for a perfect "normal" slide or neg,  vs. 25 secs for a thin original, and 40 secs for a dense original.

 

With some of my old dense negs, my Nikon, that scans a normal 35mm image in 30 secs. or so, can take up to 90 secs to allow time to "penetrate" the film's density.

 

Additionally, contrast comes from "curves" applied to the raw scan data, not exposure. A scanner will have a default curve that it can vary based on original image density and contrast, but can also be varied by user inputs.

 

Vuescan has curve controls, although they are not as "intuitive" as, say, Adobe's, where one can just grab (many) points on the curve and drag them around. In Vuescan, you have to adjust the output at the 25% and 75% points on the brightness scale, by numerical entries (e.g. set default 75% to "65%"), and move just those two points or values left or right to set a "curve" that connects 0%, the two adjustable points, and 100%, with various "S" or "U" or "C" shapes. Or leave it "linear" - a straight line - output is 1:1 with input.

_________________

 

My own scanning technique is this (Vuescan on a film-capable flatbed). Based on the wonderful fact that Vuescan can output .dng files. Processing the .DNG as a "raw" file in ACR is so much more flexible and easy and "visual" than most scanner adjustments.

 

- I use Vuescan's "autolevels" exposure mode, with clipping set to zero. Default "linear" Vuescan curve. Output as 16-bit RGB .DNG from a "B&W negative."

- I scan 6x6 negs with the black border included, and adjust "exposure" manually to give me as close to a pure black ("0,0,0") as possible in the blank areas outside the image area (film base plus fog). That can be anywhere from exposure 0.60 to exposure 1.00, depending on film type/base density/fog. But most of the time is just left at 0.7 or so. (I mostly stick with TMax 400/100 and HC-110 - very little variation in base/fog density from roll to roll.)

 

I ignore the white brightness point, at this stage. So long as it doesn't clip (and it doesn't, with negs), anything is fine.

 

- For color negs, I will also adjust the separate RGB exposures to get as close to a "correct, gray is gray" color balance - my scanner tends to make the raw image blue/cyan, so my final adjustment is often about red 1.00, green 0.90, blue 0.67.

- That gives me a scan with a "full" tonal range - everything in the actual film gets recorded with a tonal value or values other than zero or 255 (unless I blew the exposure/development and have "blank" shadows somewhere. ). I assume that is what 250swb means by a "low-contrast scan" - because that is usually the case. "Maximum dynamic range" = "low-contrast," on a linear basis.

- Those are the only scanning adjustments I make

- I do all other adjustments with Adobe Camera Raw, just as with any "raw" .DNG.

- Set blacks to just start to clip into the darkest parts of the actual image (the black borders go black long before that, if my camera exposure was correct). About 2% of the actual picture area showing as pure black, usually (foggy days, not so much).

- Then adjust "exposure, shadows, highlights, whites, contrast, curves, etc." to get the picture I want. And WB, if a color picture.

 

The steps look like this

 

a - scan as first imported into Adobe Camera Raw - option key held down to show the "clipped" blacks straight from the scanner.

b - as in a, but with the ACR blacks slider (w/ option/alt key held down) moved left to clip about 2% of the image blacks as well as the film base "border".

c - As in b, but without the option/alt key held down (shows all default scanned tones, not just black clipping)

d - after final adjustments in ACR (pretty much like LR's controls), of whites, contrast, exposure, shadows, highlights, curves, etc. etc.

 

Welcome, dear visitor! As registered member you'd see an image here…

Simply register for free here – We are always happy to welcome new members!

[Guest]

Hello,

 

if I understand correctly what you describe, you are using VueScan. VueScan adjusts the exposure time using the prescan, if hardware permits, which is the case for a Nikon.

 

However, you are mixing hardware arguments with software procedures. I can only repeat the argument, that the only thing that counts is the raw data as delivered by the scanner firmware. Even that is already processed by a matrix transformation in most cases (see Hunt 2004m page 556). The dynamic range can be adjusted by exposure time (in case of my scanner) or by lamp brightness (in case of Nikon, as far as I know). It is only these adjustments that determine the range of values delivered by the scanner (dynamic range). It is software, if you set white and black point -- either in the scanning software or in an external editor. In CCDs there is no such thing as an amplifier. You can change the conversion factor from electrons to numerical value, but this is not done in scanners since you would need new calibrations if this is done, which does not happen scanners.

 

Please note, that if the exposure is change you will need a new ICC-calibration for perfect results. This is completely ignored by VueScan, changing the exposure from image to image via the prescan!

 

Hermann-Josef

[Stealth3kpl]

 

My own scanning technique is this (Vuescan on a film-capable flatbed). Based on the wonderful fact that Vuescan can output .dng files. Processing the .DNG as a "raw" file in ACR is so much more flexible and easy and "visual" than most scanner adjustments.

 

- I use Vuescan's "autolevels" exposure mode, with clipping set to zero. Default "linear" Vuescan curve. Output as 16-bit RGB .DNG from a "B&W negative."

- I scan 6x6 negs with the black border included, and adjust "exposure" manually to give me as close to a pure black ("0,0,0") as possible in the blank areas outside the image area (film base plus fog). That can be anywhere from exposure 0.60 to exposure 1.00, depending on film type/base density/fog. But most of the time is just left at 0.7 or so. (I mostly stick with TMax 400/100 and HC-110 - very little variation in base/fog density from roll to roll.)

 

 

 

Thanks for this Andy. I think Vuescan has a different UI for different scanners.

For Plustek, with Vuescan in Professional mode under the Input tab, there is no "Exposure" adjustment. There is an overall "Brightness" adjustment and brightness adjustments for each RGB channel, Black and white point adjustments for each RGB channel, and the curve adjustments you mention, all under the "Color" tab, but these seem to affect the output information only rather than input via another scan.

Pete

[Guest]

Hello Pete,

 

you are absolutely correct. It looks like the Plustek is not able to adjust the exposure level. In this case and if you are going to work on the image anyhow in e.g. PhotoShop, I would set the white point to 255 or 65535 (for 8bit and 16bit rsp.) and the black point at 0. This way you avoid clipping by the scan software for all cases. Optimization can then be done in PS. This procedure should be equivalent to saving the scan in a raw format. This is not to be confused with raw formats in digital cameras. The raw scan formats should store the data as they come from the scanner. Albeit VueScan does more to them (e.g. it inverts color negatives, optionally removes dust and scratches …) so it is no longer raw data in the strict sense (this would be the data coming from the CCD).

 

Hermann-Josef

[Advertisement]

Hallo,

Look here: Interesting Leica gear