Color or grayscale for scanning B&W film?

[dct]

It was many years ago when I scanned myself at home (using Plustek 7200i and a hp SJ5590). I did a check and all were RGB 24bit.

Same for the Noritsu QSS scans I use in the past few years: The service provider delivers same 24bit color channel JPEG files for b&w too.

 

[Bob Michaels]

If I am understanding you correctly, you really do not have a grayscale file but an RGB file with each channel desaturated. A grayscale file will always have the color space GG 1.8 or GG 2.2 while a color file (even if desaturated) would be something like RGB or sRGB. You cannot adjust the color tint of a grayscale file as it has no color attributes. Hence the file size is much smaller.

Don't know about Silverfast but Vuescan's auto setting to convert to grayscale uses mostly the green channel although which channel one uses is user selectable in the "make gray from" setting.

But the bottom line is always if it works for you, don't change it.

...........................
I can adjust RBG separately in the Histogram if needed..Even for the B&W negatives. Not sure what effects in may have, since there is only a gray scale.
Not sure in 16bit gray-scale what channels are used though.

 

[sevo]

It really depends on the scanner. With the CS-9000 and its nominal 16bit/channel, I am confident that it delivers more than 12 bit per channel above noise, so scanning a single channel to black and white is fine, and the tiny benefits of a intermediate 48 bit colour stage are not worth the extra postprocessing time. On 8-bit and 12-bit scanners or noisier members of the 14/16 bit family, scanning in full colour and mixing down the colour channels to 16bit black and white will usually deliver better results.

 

[DNG]

Yes, that's right, with B&W, the RBG channel hold no color value, But, By scanning 48>24bit vs 16>8 bit, the 48>24bit has a bit more starting contrast. In fact, the 48bit HDR setting looks the same as the 48>24bit scan a far as the starting Dynamic Range.

If I am understanding you correctly, you really do not have a grayscale file but an RGB file with each channel desaturated. A grayscale file will always have the color space GG 1.8 or GG 2.2 while a color file (even if desaturated) would be something like RGB or sRGB. You cannot adjust the color tint of a grayscale file as it has no color attributes. Hence the file size is much smaller.

Don't know about Silverfast but Vuescan's auto setting to convert to grayscale uses mostly the green channel although which channel one uses is user selectable in the "make gray from" setting.

But the bottom line is always if it works for you, don't change it.

 

[Chriscrawfordphoto]

Grayscale only. The idea that scanning black and white film in RGB somehow gives smoother tonality or wider dynamic range is nothing more than a myth started by people who don't know any better, and after years of it circulating online, a lot of people blindly accept it. I've thoroughly tested this on my Nikon 8000ED scanner and scanning in RGN gives you nothing more than a file that takes up three times the disk space and requires more computer resources to edit.

 

[sevo]

Grayscale only. The idea that scanning black and white film in RGB somehow gives smoother tonality or wider dynamic range is nothing more than a myth started by people who don't know any better, and after years of it circulating online, a lot of people blindly accept it.

Well, there are some valid reasons behind it. Assuming a scanner whose low curve end is smothered in noise (which is generally true), you'll get something more than a half and less than one bit extra out of averaging three channels rather than directly using one channel. This is no matter of colour - scanning one channel in three passes and averaging will get you the same results (however it will amplify the defects of that one sensor line, if any). And using the RGB file directly without averaging into grayscale will give you colour noise rather than extra depth - so the notion that storing to RGB is somehow beneficial is wrong (at least if your software allows for writing to a better than 8-bit grayscale file).

 

[DNG]

Grayscale only. The idea that scanning black and white film in RGB somehow gives smoother tonality or wider dynamic range is nothing more than a myth started by people who don't know any better, and after years of it circulating online, a lot of people blindly accept it. I've thoroughly tested this on my Nikon 8000ED scanner and scanning in RGN gives you nothing more than a file that takes up three times the disk space and requires more computer resources to edit.

Not to start a fight....
But, I see a real difference in the histogram width and Peaks from a 48>24bit and a 16>8 bit scan

Now, the DR may not be much, but, it is very plain to see, the 48>24 bit scan of a B&W negative has a wider histogram, it may even clip a few points off the shadows. Looking at the top Histogram.

The 16>8 bit shows a histogram that has no clipping...

48>24 bit Capture

16>8 bit Capture

 

[Bob Michaels]

............. you'll get something more than a half and less than one bit extra out of averaging three channels rather than directly using one channel. ..................

One more time: you are going to average 3 channels no matter what you do. Your scanner is going to scan a red channel, a green channel and a blue channel, even if you are scanning a b&w neg. This is inherent in all scanners and there is no way you can control it. This unmanipulated data is going to be transferred back to the CPU where it is processed by the scanner driver.

If outputting a grayscale file, your scanner driver is going to combine these three channels for you. And, almost all seem to be an excellent job of it. Or, you can output the three very similar channels in an RGB file and average them yourself.

Separate issue:
I do see advantages to a "high bit" file be it a 48 bit RGB file or 16 bit grayscale file. Remember the 48 bit file is 3 channels of 16 bit data. The 16 bit channel will contain 65,536 data points or shades vs. 256 for an 8 bit channel. This is what makes high bit files look smoother in the transitions.

 

[Chriscrawfordphoto]

Not to start a fight....
But, I see a real difference in the histogram width and Peaks from a 48>24bit and a 16>8 bit scan

Now, the DR may not be much, but, it is very plain to see, the 48>24 bit scan of a B&W negative has a wider histogram, it may even clip a few points off the shadows. Looking at the top Histogram.

The 16>8 bit shows a histogram that has no clipping...

48>24 bit Capture

16>8 bit Capture

That's being done by the software. You will get a different result with different scan software, like Vuescan. In any case, your examples are showing worse scans with color than with greyscale (because your software is choosing to clip the shadows on the color scan).

 

[DNG]

That's being done by the software. You will get a different result with different scan software, like Vuescan. In any case, your examples are showing worse scans with color than with greyscale (because your software is choosing to clip the shadows on the color scan).

Yes, I saw that also... I am scanning a roll as I watch this thread, I am now scanning the balance in 16>8 bit gray scale.. This topic has adjusted my workflow for B&W scanning.

 

[Johnmcd]

16 bit grey scale at the highest optical resolution on my Coolscan V. When I tested that against the colour and select a channel approach I couldn't see an advantage in either and so chose the one that gave me the smallest files.

I once asked Ed Hamrick of Vuescan fame if he could have a mode in the software that scanned negatives as positives and then inverted the scan. His response was that was what the scanner did anyway so there was no point.

When I first got my Scan Multi Pro I always scanned as a colour positive and then inverted in PS. Using the Minolta software, this was much better than the B/W setting.

I then purchased Vuescan as the Minolta software didn't work with Win7. With Vuescan I could see no difference, so I just scan in 16bit greyscale. Now I know why

 

[venchka]

I obviously know nothing. That explains why I have banned myself from these discussions.
If you would like to know a thing or two about scanning, GOOGLE "Ken Lee Scanning Tips." Ken does know a thing or three about scanning.

Wayne

 

[sevo]

One more time: you are going to average 3 channels no matter what you do. Your scanner is going to scan a red channel, a green channel and a blue channel, even if you are scanning a b&w neg. This is inherent in all scanners

No. There are scanners that very obviously scan only one channel in black and white mode - three pass or three colour light scanners (like my old Epson Slidescan) that do only one pass or use only one light source in B&W cannot scan three channels for obvious reasons. And whether the software discards two channels or mixes three is entirely up to its maker and purpose - scan software will often do the former when you scan a black and white original to a black and white destination (good software will even let you select the channel used), and must do the latter when registering the original as colour and a black and white destination.

 

[Bob Michaels]

No. There are scanners that very obviously scan only one channel in black and white mode - three pass or three colour light scanners (like my old Epson Slidescan) that do only one pass or use only one light source in B&W cannot scan three channels for obvious reasons. And whether the software discards two channels or mixes three is entirely up to its maker and purpose - scan software will often do the former when you scan a black and white original to a black and white destination (good software will even let you select the channel used), and must do the latter when registering the original as colour and a black and white destination.

Sevo: Yes, you very well may be correct that there are some old scanners that only use one of the color channels to output a b&w image rather than mixing all three. Hopefully I sufficiently addressed the situation of the scanner software converting the color data to monochrome in my earlier post.

Does the old Epson Slidescan actually do 3 separate passes for color, one R, one G, and one B for a color scan but only one pass (probably green) for a monochrome scan? I would think the registration of the 3 passes would be difficult. Or, does it just illuminate one of the 3 color light sources?

My key point is that the scanner itself has no dedicated monochrome scan mode, that it all is in software somewhere along the process. I guess someone could make a dedicated monochrome film scanner similar to a M9 monochrome but then it would no longer do color.

Putting it another way: all the scanner itself does is shine a light through the film, read the intensity on a receptor and output that basic unprocessed data back to the scanner driver in the CPU. It does not know if it is scanning a positive or negative, color or monochrome. That is all in software.

 

[sevo]

Does the old Epson Slidescan actually do 3 separate passes for color, one R, one G, and one B for a color scan but only one pass (probably green) for a monochrome scan? I would think the registration of the 3 passes would be difficult. Or, does it just illuminate one of the 3 color light sources?

It has a three colour diode light that oscillates between R, G and B in colour mode, and stays green in black and white.

My key point is that the scanner itself has no dedicated monochrome scan mode, that it all is in software somewhere along the process.

That is probably true for all current or recent scanners - there is not much point in doing the processing at the scanner side now that even the most humble computers will do just about all 2d still image processing tasks in realtime - but earlier scanners did quite a bit of the processing in the scanner hardware, to reduce the volume that had to be transferred and relieve the CPU.

 

[kuzano]

Color!!!

Color!!!

Presumably, you are going to post process the image.

If so, you have a lot more effect with your color filters in Post, if you are using the color information in the scan file.

How is that so hard to grasp?

Common Sense... without the discussions of bits, bytes, color pallets, etc

I once asked a friend if I should shoot in monochrome with my digital camera, or color when the end result was going to be B/W. He simply said, If you are going to take the trouble to drive out in the woods, find a subject (landscape), set up and take the shot, why would you drive back home with only part of the information on your memory card... Same with scanning.

He is an excellent film landscape and still subject photographer. He also tells me that he shoots mostly color film for the increased information when the file is scanned.

He further says, that the only black/white he shoots is when he wants an image where a particular film emulsion renders a unique, desirable quality.

It's so logical, I truly have trouble with those people who say digital for color and B/W film for monochrome. It's your job to make the B/W desirable or render the quality you are looking for in an image. Unless you are still film and "wet" process and print all the way, you must make the digital decision at some point.....

I say scan and scan in color.

 

[sevo]

Presumably, you are going to post process the image.

If so, you have a lot more effect with your color filters in Post,

Uh, we were talking about black and white negatives. No possible effect with colour filters there - the only question is whether merging the channels will deliver a better result than a single channel, and if so, which software (and which command inside the software) delivers the best results.

 

[Particular]

One more time: you are going to average 3 channels no matter what you do. Your scanner is going to scan a red channel, a green channel and a blue channel, even if you are scanning a b&w neg. This is inherent in all scanners and there is no way you can control it. This unmanipulated data is going to be transferred back to the CPU where it is processed by the scanner driver.

There is a way to control it! When you scan 16bit b+w in VueScan, there is the option to; Make gray from: Red, Green, Blue or Infrared. You can choose which channel has the best output.
I have done test scans from some of my own negatives with each channel to see the difference. The difference was very subtle, and really only was visible in the look of the grain when viewed at 100%. The blue was harder looking, and red was softer looking with green somewhere in between. Infrared was even softer, perhaps because the focus shifts with the color wavelength. I decided that green was the compromise that would work best in most situations.

 

[Bob Michaels]

CNNY: you are talking about what VueScan does with the unmanipulated data it receives from the printer. I understand that having used VueScan for 10 years. VueScan cannot control the way your scanner scans that is built into the hardware and firmware.

The key point is that the data the scanner extracts from the film is always the same, regardless of any settings, and inherent in the scanner. The difference is how the software (VueScan, EsponScan, NikonScan) manipulates that data in creating the file it exports.

There is a way to control it! When you scan 16bit b+w in VueScan, there is the option to; Make gray from: Red, Green, Blue or Infrared. You can choose which channel has the best output.
I have done test scans from some of my own negatives with each channel to see the difference. The difference was very subtle, and really only was visible in the look of the grain when viewed at 100%. The blue was harder looking, and red was softer looking with green somewhere in between. Infrared was even softer, perhaps because the focus shifts with the color wavelength. I decided that green was the compromise that would work best in most situations.

 

[venchka]

Yo! Folks. Chill. If it works for you. If it looks good. Just do it!
Personally, I use Epson Scan, 16 bit gray scale, black & white point & gamma set to my taste, output set to 0 & 248, TIFF output, Lightroom processing. Done. Works for me.
YMMV.

Wayne