scanning b/w negs

[northeast16th]

So I'm scanning black and white negs, and this seems to be the best way to do it. Can anyone confirm that they have had success the same way?

I scan in 16bit monochrome as a transparency and then invert in PS. I doesn't look so good straight from the scanner, but I can really tweak it without losing too much info if I need to. I have also tried making adjustments in SilverFast with 16 -> 8 bit (whatever that notation means), but the histogram never quite has the range that 16 bit has even if it looks much better straight from the scanner.

This seems to be what works best, but I haven't made prints yet from the scans.

Thanks in advance for any feedback.

 

[ElGato]

That is what I do with SilverFast and my Plustek as well. The canned curves for various film types do not work too well for me, so I just do the "RAW" scan and fix it up in Photoshop. I have actions set up to invert, set color profile to gray gamma 2.2, set levels to 0.05% clipping on each end, and apply a curve that generally works for the type of film. I save these off as jpegs, and keep the original TIFF untouched. When I've finished a batch of scans, I'll go through the jpegs and figure out which ones I want to process more carefully, and start back from the original TIFF.

 

[Chriscrawfordphoto]

Yep, you've pretty much got it right. We were just discussing this in another thread!

 

[Tim Gray]

Your mileage may vary, but two things:
1. invert in photoshop is not really what you need to do for negatives. It's 1-x (where x is the rgb of the image) whereas you really want 1/x.
2. gamma corrections in photoshop get funky much beyond 1.5.

These two things aren't the end of the world with B&W scans, since you don't have to deal with colors, saturation, etc. They really get messed up by the above two things. With B&W, you can just do a more extreme curve adjustment. So, just some things to keep in mind.

You can test number two by taking a gradient and putting two levels adjustment layers overtop of it. Put middle slider on the first at .5 and the second at 2. They should cancel each other out, but you don't get the original gradient back.

I personally think you are better off scanning as a negative and NOT inverting in photoshop.

 

[Bob Michaels]

Keep in mind that your scanner does not "scan" as a positive or negative. It merely records the light passing through a piece of film at many points. The scanner itself does not know, nor care, if it is scanning a positive or negative or the original scene. That data is passed along to the scanner driver software (Nikonscan, Minoltascan, Vuescan, Silverfast) that interprets it in your computer's CPU and configures the output file as a positive or negative. Same situation applies to RGB vs monochrome.

But if having your scanner driver output the file one way (positive vs. negative) and you inverting it in an image editor works for you, do what you think works. That was recommended years ago when Nikonscan and Minoltascan made some funky behind the scenes adjustments to negatives that they did not make for positives.

Also remember that the initial scan file that looks the best is actually not the best. The best scan file is what gives you the best final output. That means the best scan will probably look flat because it is passing 100% of the data for you to adjust as you want in your image editor.

 

[ElGato]

That 1/x inversion story is interesting. It explains why I have to make large curves adjustments in Photoshop. Unfortunately, from what I can determine, Silverfast SE will not spit out a pre-inverted 16 bit grayscale image.

 

[kzphoto]

Scan in as a color neg and use the channel mixer in photoshop to make your image monochrome. You'll capture more information.

http://www.flickr.com/photos/xbleh/sets/72157614645496177/

 

[Chriscrawfordphoto]

Scan in as a color neg and use the channel mixer in photoshop to make your image monochrome. You'll capture more information.

http://www.flickr.com/photos/xbleh/sets/72157614645496177/

That only works if the negative you scan actually IS a color neg. Scanning a BW neg in color does nothing for you and the channel mixer trick doesn't work on an image that starts out BW.

 

[mfogiel]

There is no need to scan in RGB unless the negative has a stain ( from pyro developers). Also, scanning as positive is only necessary if there is clipping when you scan as negative, Develop shorter, and the problem will be solved.

 

[Tim Gray]

Another interesting note: One of the RGB channels might be sharper than the others. I scan with the B channel when I do B&W scans because it is noticeable sharper than the other two.

 

[Bob Michaels]

Scan in as a color neg and use the channel mixer in photoshop to make your image monochrome. You'll capture more information.

Again, all scanners do the actual "scan" using the red channel, the green channel and the blue channel even when scanning a b&w neg. The channels are very similar using a b&w neg as Chris points out. It is simply a matter of you choosing to let the scanner driver software combine the 3 similar channels and output a greyscale file or having the software output all 3 similar channels and you converting them to greyscale with your image editor.

I keep bringing this up only because many continue to think that the raw data captured by the scanner differs depending on the positive/negative, RGB/monochrome, 16bit/8bit settings. It just ain't so. The initial data read by the scanner is consistent. It is only what the software does in the CPU with that data that differs.

I do not know if Nikonscan, Minoltascan or Silverfast scanner driver software do some funky things with the file before they output it. I only know that Vuescan does not. Using Vuescan it makes no difference if you output the file as a negative or positive, RGB or monochrome.

 

[Bob Michaels]

Another interesting note: One of the RGB channels might be sharper than the others. I scan with the B channel when I do B&W scans because it is noticeable sharper than the other two.

I note where current versions of Vuescan give you the choice of making the greyscale file from the R channel, the G channel, the B channel or "auto". I honestly do not know how "auto" does the conversion but I have always used it.