Black and White Scanning Tips

In case this topic has not been beaten to death, could some of you

scanning mavens share with the rest of us your secrets for getting the

most out of your black & white negatives. I'm particularly interested

in better understanding the advanced settings on my scanning software

in order to tease out shadow detail while clipping as few highlights

as possible. I'm now playing with both Nikon Scan 4 and Vue Scan on a

Super Coolscan 5000.

 

In addition to better understanding some of the more advanced

settings, I'd like to know what you think about:

 

(1) using multi-pass/sampling (4x, 8x or 16x);

 

(2) 16 bit scans (so that there is more meat to work on when fiddling

with Curves or Levels);

 

(3) scanning negatives as positives and then inverting and bumping up

contrast in PhotoShop; and

 

(4) scanning b&w negatives as color negatives and desaturating in

Channel Mixer as ways of getting better quality scans.

 

(5) what does fiddling with the "black point" do?

 

(6) any other tips?

 

BTW, if this topic is well-covered in the archives, kindly post a link.

 

Thanks

Marc,

 

The attached .ini file are the settings I use with Vuescan for traditional B&W films (regardless of which one). The image scans a little flat but is easily fixed in PS.

 

1)

The multi-scan is set at 16x and the argument behind this is to get more accurate data in the first place. This scanner may have the abiltity to represent a channel as 16-bits and translate this to a Dmax of 4.8 but in reality it wont come close to achieving this level of accurate data. From sampling theory, every doubling of the sample rate should give you an extra bit of information. So if the CCD is capturing 8~10 bits of real data from a single pass then 2x = 9~11, 4x = 10~12, 8x=11~13, 16x = 12~14 is possible although many will challenge being able to see a difference. Whatever... this is just what I do and why; it takes longer to scan and if its not your thing just reduce the multipass or turn it off - your choice!

 

2)

Yes... especially important for B&W work which is more tonally focused than colour and often is subject to manipulations

 

3)

This is not necessary on the Nikon, especially if using Vuescan. I just dont use Nikonscan but I hear that the initial problems in capturing detail that scanning as a positive was intended to corrrect may have been addressed in the later versions.

 

4)

With the Nikon and using Vuescan, if you select scan as B&W neg it IS scanned as a 3-channel RGB anyway. Its obviously desaturated somewhere in the process but I have never found it necessary to make it an active step to get a good scan.

 

5)

Change how much "black" you have in the image which in turn relates to shadow detail. Traditional B&W prints dont run a continuous tonal range from 0 to 255, they tend to hit 100% black at around the 85% ~ 90% mark. This helps gives them that "black" and "white" look as against a smooth greyscale look. Digital has a better ability to display a tonal range well into the shadows (and the same on the opposite end) so if you are only hitting "black" at 0 (100%) then the image can tend to look more of a smooth greyscale rather than what has been traditionally seen in B&W prints.

The setting in the .ini file attached has cliping set at 0.01%, this can be changed as require latter in PS with levels settings.

 

6)

I often find that better tonal work in PS can be achieved with local manipulation versus global (ie, dodging and burning) after the initial global adjustments. I still use PS7 and in 16-bit mode achieve this by applying a global curves adjustment focusing only on how it affects the local part under consideration then by marking this as a history state and returning to the image state before before the curves adjustment, I can use the history brush tool on about 20% opacity to apply dodging and burning to the appropriate areas.

 

Sharpening in B&W scans needs individual care (I find more so than colour). I use an edge mask first (there are some ok actions for achieving this in PS on www.pinkheadedbug.com. These actions wont work in 16-bit mode so ...within the action turn off everything after it has made the edge selection, make an image duplicate, convert it to 8-bit then run the action; with the 8-bit image selected go under [select]. choose [save Selection...] call it anything and save; return to the original go back and choose [Load Selection...] find the name you called it and load, you should see the same marching ants. Now run the USM. I find for B&W settings much more than 80 are difficult with a 1 pixel radius and a threshold of 5. I examine the image at 50% and especially in OOF areas the USM isnt nice ...so use the history brush and either paint out or in the sharpening as desired.

 

hope this is of some help...

 

Craig

Following relates to both b&w traditional negs and chromes using Vuescan in Minolta MultiPro.

 

1) I have never seen an improvement from multi-pass scanning so don't use it.

 

2) I always scan in 16 bit as it sometimes does make a difference eliminating posterization caused by radical contrast curves.

 

3) & 4) I always scan b&w negs as negs and chromes as RGB positives. The original data the scanner lens reads is always the same and the Vuescan algorithms to convert seem as effective as PS.

 

5) I always check the black point and white point by looking at the histogram to insure there is no clipping. I just try to capture all the data in the scan and no more. The black point just sets where the data stops being included in the file.

 

6) I very occasionally use the "long exposure pass" option when a neg or chrome has needs the range expanded. It seems to help picking up highlight detail (the dense part) of negs or shadows in chromes. Maybe it's the other way around, can't remember as I use it so infrequently. This is the one time I will try to scan a neg as a positive to see if it helps.

 

I use the Vuescan default settings 98% of the time. My theory is that the scan is nothing more than capturing as much as possible to adjust in PS. I only bother to insure there are no clipped end points of the histogram.

 

My workflow for b&w is simple:

 

1) set film type (not critical, I frequently use XP2 even with silver

film as it works)

 

2) Preview scan at final desired resolution. Vuescan is set to eject film which I file away at his point.

 

3) adjust cropping if necessary

 

4) check histogram to insure there is no clipping. Adjust endpoints if necessary

 

5) click to save file and clear memory. Begin next scan

 

Think of it as scanning a neg or chrome as having only two distinct steps:

 

Step 1) scanner shines light through film and internally holds this data. This is always the same no matter if it's a neg or positive, color or b&w, cropped or not. This is like RAW data captured by a digitial camera. (I think Vuescan long exposure pass does two scans here with different analog gain adjustments)

 

Step 2) scanner software uses the CPU to adjust this data to show you on screen and in the file you save. This is where the raw scan data is converted to either a neg or a positive image, color is converted to grayscale, part of the file is deleted for cropping, endpoints are set, colorbalance or contrast is adjusted, dust reduction (if any) is applied, sharpening (if any) is applied, and all the other adjustments.

 

It begins to make more sense to me to analyze the scanning process in terms of these two distinct steps. You then realize the 2nd step is critical only not losing any desired information.

Craig, using your ini file sets Vuescan to Color Neg and Fuji NHG profile....is that right? Also White Balance set at 1 (serious highlight clipping) and output JPEG (no 16-bit then of course)? Something's not right....

Clarification to my previous post: When I said "Step 1) scanner shines light through film and internally holds this data" could be misunderstood.

 

Actually the scanner digitizes through firmware and sends this raw data it to the CPU. Everything past this point takes place in the computer, not in the scanner. In fact, you could unplug the scanner unless it makes the driver shut down.

Response to:

 

(1) never bother

 

(2) yes

 

(3) no, see Vuescan workflow below

 

(4) no, d.o.

 

(5) clip black

 

(6) Try Vuescan scan-from-disk

 

Here are my last default settings for Vuescan. Note, looking at the results a year or so later, I'm thinking the contrast really needs a push, or something! Also, the logic with "lock film base color", which was just manually set, was to avoid any intial clipping of the left end of the histogram. Not sure if it's really needed.

 

Anyway, this might serve as a starting point, if you try Vuescan:

 

++++

 

Input|Media type: B/W negative

 

Input|Bits per pixel: 48 bit RGB

 

Input|Lock film base color: ticked, settings per list below

 

Crop|Buffer%: 0

 

(if you have some "hot" spots on the edges, you can raise the buffer %, to intentionally blow them out)

 

Color|B/W Vendor: Kodak

 

Color|B/W brand: TMax400

 

Color|B/W type: D76 CI:.55

 

(starting point, good for most. If you have blowout, you can raise this number to quell it. Or if your picture seems very flat, lower the number)

 

Color|B/W curve: manual

 

Color|black point: 0.02%

 

Color|white point: 0.02%

 

Color|brightness: 0.7

 

Output|Raw file type: Auto (48 bit rgb)

 

Output|Tiff file type: 8 bit grey

 

++++

 

Just a comment on the posted Vuescan ini file. It's very difficult to read in that format. Most of it is in the esoteric language intended for communicating with the program.

<i>"Just a comment on the posted Vuescan ini file. It's very difficult to read in that format. Most of it is in the esoteric language intended for communicating with the program."</i><p>I didn't "read" it. I actually loaded that ini file in Vuescan.

1) If your scanner can register passes/samples accurately, sure, it reduces any CCD noise

present. I've never had a scanner that could do this without wild registration problems, but

your Nikon could easily be better.

 

2) Any time you plan big PS moves scan 16-bit. If you're scanning 16-bit don't do any

adjustments in the scanner software, do it all in PS.

 

3) No, always as negs, see below

 

4) Depends on the scanner. On at least some scanners, Vuescan handles the presumed

color mask by changing the analog exposure on each channel, not just twiddling bits.

Having channels with different analog exposure can save negs with a tonal scale that just

won't fit otherwise. I've found it's the only way short of a drum to get most of the data of

pushed TMY, for instance. Since your Nikon is reputed to have even worse trouble with

dense silver negs than my Polaroid 120 I had I'd greatly recommend trying this.

 

5) The same thing as the top left slider in the PS levels dialog. I never bother--if you're

scanning 16-bit it's easier--and more accurate--to just deal with this in PS, where you've

got an undo. If you blow the blackpoint during a scan you have to rescan.

 

6) Underdevelop. CCD scanners in general, and collimated-light-source Nikons in

particular, have trouble with dense areas in silver negs. TMY especially can hold detail in

incredibly dense highlights, but no CCD scanner on earth is going to be able to dig it out.

fyi Nikon doesn't have problems with B&W, the issue is with Nikonscan, not the scanner.

 

Simply switching to Vuescan is magical.

 

Chatter about different light sources (eg old Minolta Vs new Minolta) is urban legend...the new Minolta and Nikon Coolscan do beautifully with B&W with appropriate straightforward use... 5-10X faster than the old hot Minolta, the new are cool: the old pops film like a slide projector, reducing sharpness: the new, being cool, don't. IMO

Craig, I may be wrong, but I think your point No. 5 is wrong. Digital has less ability than

negative film to extend tonal range into highlights and shadows. In other words shadows

block and highlites blow out in a narrower tonal scale than negative film. Digital is more

akin to shooting chromes in that regard.

Hi Richard. I know what the file is for. I'm just thinking, to communicate settings, it would be better to describe them in ordinary terms.

In my experience with 20D, it behaves even a little worse than a typical slide film, ie: the highlights blow very easily.

Yes, Mendel. I just thought maybe you thought I had only opened the file and read through it that way. Either way I suspect it did not indeed "translate" over for some reason....

Hi Craig,

 

Thanks for the Vuescan .ini file. I've tried it on a few b&w negatives and if works great. I've got a few more negatives to try and if they work out i'll keep the file.

 

Like you've said, "The image scans a little flat but is easily fixed in PS". I like that as it tends to preserve both highlight and shadows, as once data is lost at the scanner level, there's no way to get it back.

 

Cheers.

Roger: you said "Vuescan handles the presumed color mask by changing the analog exposure on each channel, not just twiddling bits"

 

This is very interesting and new data to me. My presumption has been that the scanner firmware always passes the same raw scan data to the computer where the scanner driver manipulates it (i.e. positive or negative and everything else) before you save the file. If it changes the analog gain, that's a whole different story.

 

Can you give any references or other info when the analog exposure is tweaked for each channel? Do you think there is any other situations where the driver actualy changes the raw scan process rather than "didling bits" (good term BTW)

 

Does it do this in the preview scan as well? Important as I never do a final scan, just save the adjusted preview. This does seems to work for me.

 

FWIW, I have my MultiPro calibrated with an IT8 target for Fuji Astia which is about the only color emulsion I shoot. And b&w just seems to always work well no matter how I do it so long as I don't clip. So my questions may be academic. But I do like to understand the underlying fundamentals rather than relying on black magic.

Richard,

<p>

<i>using your ini file sets Vuescan to Color Neg and Fuji NHG profile....is that right? Also White Balance set at 1 (serious highlight clipping) and output JPEG (no 16-bit then of course)? Something's not right....</i>

<p>

Not sure whats going on for you with the film type but - <i>looking at this I can see how this might be a version of Vuescan issue as it appears to reference which item number in the list of options and as options are slightly changed across versions, Im running 8.2.30 ???</i> but i went back through and checked and for me this is:<p>

NegativeVendor=3<br>

BWBrand=18<br>

BWType=2

<p>

...which translates into: Kodak / TMAX-400 / D=76 CI = .55

<p>

I hadnt found to have had much trouble with the white point settings.

<p>

[Output]<br>

TIFFFile=1<br>

TIFFFileType=4<br>

TIFFCompression=0<br>

TIFFProfile=0<br>

JPEGFile=0<br>

<p>

This should set TIFF output.

<p>

Craig

Craig-

 

How do you go about loading an "ini" file in VueScan? Do I first need to save it to my desk top or do I print it and then type in the settings?

 

Also, can VueScan be opened up from PhotoShop like other scanning programs, or msut it always be a "stand alone" program?

Marc: Vuescan lacks the TWAIN interface to open directly from Photoshop. But running it as a standalone program is much better.

 

If you have a fast enough CPU, enough memory, and good attention span; you can multitask and simultaneously:

1) be scanning a neg

2) working up the previous neg in Photoshop

3) be printing the neg previous to that.

Marc, ini files are loaded in the file pulldown. Same place for saving them, or loading defaults.

 

Still, I kind of "bristle" over this. Just loading someone's ini, you get no idea what all the settings are. There may be settings in there you do not want/need.

Bob:

<br><br>

(From Vuescan Users Guide, "Scanning Black/White Negative Film")

<br><br>

<i>On most scanners, setting "Input|Media" to "Color negative" will increase the green

exposure time by 2.5x and the blue exposure time by 3.5x. This results in adjusting for

the green and blue absorption by the orange mask of the film. If the film doesn't have an

orange mask, then using "Color negative" will result in a raw scan file that looks very, very

cyan.

</i><br><br>

As far as I know this is the only place it does this.

<br><br>

At one point I tried very hard to convince him to allow a user settable exposure difference

between channels, even offered to pay him, but got nowhere.

<br><br>

There a slight drawback to this. On some scanners you're using the CCD at an exposure

length other than what it was calibrated for, leading to possible pixel-column-bias. (That

is, your image has very faint stripes) This mostly isn't an issue unless you're making

absolutely huge levels/curves moves on a very difficult negative, but I have had it be a

nuisance once or twice on a Sprintscan 120.

Roger: You seem to be right. This seems to be the only change impacting the raw data coming from the scanner firmware. So maybe it would make a difference in a color neg scanned as a positive and then inverted. But that should make no difference how you characterize a clear base neg (chrome or b&w neg) as positive or negative in the driver and subsequently invert in PS. This relates to Marc's original quesiton.

 

On the suject of my diversion (is a final scan really necessary if you use Vuescan?) Ed Hamrick says the only difference between the raw data compiled by the preview scan and the final scan is an exposure adjustment which may or may not be necessary. My experience is that it's never required with normal b&w negs or chromes. I guess that's why my technique of doing the preview at the final desired resolution, adjusting and then saving it without doing a final scan works.

<i> But that should make no difference how you characterize a clear base neg (chrome or

b&w neg) as positive or negative in the driver and subsequently invert in PS.

</i><br><br>

The point with a clear base neg is that you have channels with 1.75 stops difference in

analog exposure, and thus that much more dynamic range when combining them.

<br><br>

For a normal neg this may be irrelevant, but for a "long" neg--like the pushed TMY I used

it with--this extra headroom can make all the difference in the world.

Roger: Apparently you find that method works better than the "long exposure pass" option?

 

I haven't experimented in years but my tests showed nothing but the "long exposure pass" made any real difference. Not scanning as color vs grayscale, not scanning as positive and later inverting, or anything else.

 

But we all have to do what works for us. It's appearing there is no best way to scan film, just like there's no best way to develop it.

Bob: Sorry, I should have mentioned that my scanner, a Polaroid 120, had registration

issues that made long exposure pass not a viable option. Yeah, if that works it's probably

better, although having full control of the effect in a 3-channel file still appeals to me

more.

As long as the raw scan is saved "at scan", the rgb adjustment with color neg setting is _probably_ the only raw file adjustment. If you save the raw scan "at save", be careful: in addition to the documented extra inclusions in the raw file data (such as infrared cleaning, grain reduction, etc.), Vuescan will modify the color balance according to any custom ICC Profile you have currently set. This is contrary to the documentation, and basically a bug. Perhaps it has been fixed of late, don't know.