chatanooga
Member
I use a Nikon 9000 and scan in grey-scale. I have to assume Nikon assign the optimum channel for the scan (eg green). Creating an RGb representation of a BW negative seems a big waste of disk-space and scanning time. . If each channel's data is the same then you only need one of them; if they're different then you have some kind of uncontrolled color-cast. I would rather do any color conversion (eg PS 'photo-filter') when I've scaled the image down to print-size and retain the 'grey' image for archive.
Faintandfuzzy
Well-known
I used to use a 9000 for medium format. The channels are not optimized as I saw better noise and sharper scans using the green channel.
Now whether or not that will show in an 8x10 or 10x15....probably not.
mh2000
Well-known
no, this isn't completely true. the main benefit IMO from scanning in 16 (14) bit color and converting is that you get 3 separate channels that get averaged in the conversion to b&w process. This is similar to a multi-pass scan and reduces scanner noise... but scanning in color essentially get the multi-pass scan in a single pass. For very sharp scanners averaging out the noise also reduces the grain aliasing.
So, yes, if you got a perfect sample from a single scan it shouldn't make a difference if you scanned in b&w or color and converted, but it is not a perfect world and digital noise is a reality.
With the soft scans from my 4990 and larger formats the noise and aliasing are not the limitiing factor so I scan in b&w to save time, but in my dedicated 35mm scanners all my tests show much nicer grain level scans from color.
Vuescan offers the option to scan in color, but saves a 16-bit tiff grayscale... I'm not sure what channel mix coefficients Hamrick is using, but it seems fine. In Nikonscan I scan in color and convert in PS (mostly just convert to grayscale unless there are problems).
>> I reckon that if your target is the the same bit depth (say 16 bits), then scanning at 16 bit gray, or at 16-bits per color RGB and converting to gray afterwards, should get you the same results either way - as every pixel you scan is gray
So, yes, if you got a perfect sample from a single scan it shouldn't make a difference if you scanned in b&w or color and converted, but it is not a perfect world and digital noise is a reality.
With the soft scans from my 4990 and larger formats the noise and aliasing are not the limitiing factor so I scan in b&w to save time, but in my dedicated 35mm scanners all my tests show much nicer grain level scans from color.
Vuescan offers the option to scan in color, but saves a 16-bit tiff grayscale... I'm not sure what channel mix coefficients Hamrick is using, but it seems fine. In Nikonscan I scan in color and convert in PS (mostly just convert to grayscale unless there are problems).
>> I reckon that if your target is the the same bit depth (say 16 bits), then scanning at 16 bit gray, or at 16-bits per color RGB and converting to gray afterwards, should get you the same results either way - as every pixel you scan is gray
Robin P
Well-known
Interesting! Just by coincidence I came across this thread while waiting for my Epson 4490 to complete a scan of a 6x6 B&W neg - so I was ready set up for a quick experiment.
First I'll mention that I only have experience with C41 process B&W films and on this occasion I'm scanning Fuji Neopan 400CN (Ilford XP2) which has a purple (rather than orange) mask.
Usually I set the Epson software to Colour Neg, 16bit Greyscale 3200dpi, if the neg is well exposed I let the software "autoexpose", set the Unsharp Mask to Low and the Digital ICE to Speed.
Having done a scan at 16 bit greyscale I did a second one at 48 bit colour and compared the two side by side (after changing mode to greyscale to remove the brownish colour cast on the 48 bit file). There was no noticeable difference in tonality or sharpness but the colour scan had more apparent "noise" or grain!
For this scanner I'll stick to 16 bit greyscale for B&W.
Cheers, Robin
Faintandfuzzy
Well-known
I'm confused. You said you scanned as a color neg, 16 bit greyscale. There is no such setting as color neg is not greyscale. As well, there is no point in scanning as color and converting to greyscale in one step.....you've missed the most important part. What needs to be done from the color scan is to view the 3 color channels in Photoshop....pick which one is sharpest, with the least noise, and delete the other two.
Give that a try.
myoptic3
Well-known
That probably depends on your scanner and your image. W/ my Nikon V-ED I always scanned in gray scale because the color scan mode gave me more information, but worse images. They looked more like what a conversion from color film to B&W might look like. I now have a Minolta ScanDual III and on some images the gray scale works better. on others the color mode. It's aggravating because I have to scan in both modes to see which is better and delete the reject. But scanning isn't much fun anyway. It's just something I get through.
Robin P
Well-known
Here's the setting in Epson Scan.
Looking separately at the three colour channels, none of them is as clean or noise free as the original greyscale scan so I'll stick with that for this scanner.
Cheers, Robin
Attachments
JTK
Established
"... I printed the images at 13 x 19 and saw absolutely no differences. Maybe for color, but not for B&W."
The print tells the tale.
There MAY be some advantage to color scanning if you use a home-made staining developer (rare kind of developer)... that's what a published staining-developer authority told me, though even he hadn't found it helpful. His suggestion was to try color, then play with color channels to see if you prefer ONE of them: he guessed they might be very slightly different with a staining developer, wouldn't be with conventional developers. He personally scanned B&W as B&W.
The print tells the tale.
There MAY be some advantage to color scanning if you use a home-made staining developer (rare kind of developer)... that's what a published staining-developer authority told me, though even he hadn't found it helpful. His suggestion was to try color, then play with color channels to see if you prefer ONE of them: he guessed they might be very slightly different with a staining developer, wouldn't be with conventional developers. He personally scanned B&W as B&W.
anselwannab
Well-known
Scan Dual IV, Tmax 100 and 400. Color Positive with 2x sampling, 16 bit color. Look at the "negative" produced and change the exposure to make sure highlights/shadows not clipped. Send to PSEIII, invert, desat. Use a curves program to get tonality right. Print with QTR, because I usually get a cast with printing right from PSE.
Method seems to prefer thin negatives.
I am going to try just using the green channel, that sounds interesting and plausible.
Method seems to prefer thin negatives.
I am going to try just using the green channel, that sounds interesting and plausible.
danwilly
Established
I found that going 16 bits doesn't add anything. Eight bits grayscale will give you all you need. There is a misperception that more data is better. All you gain is a larger file size.
Faintandfuzzy
Well-known
Wow Dan....you've just turned the scanning world on its head.
In fact, you do get more with a 16 bit scan. First off, an 8 bit scan doesn't give you a completely accurate 8 bit scan....there is a margin of error. With a 16 bit scan, tonal transitions are smoother and you avoid posterization when adjusting the image to bring up the shadows or darken the highlights.
With an 8 bit scan, you have the following data available to work with:
in the first stop...
128 tonal transitions
2nd - 64
3rd - 32
4th - 16
5th - 8
6th - 4
7th - 2
So as you can see, if you are adjusting to bring out any details in shadows down a few stops, you'll have posterization issues with only 8 or 4 tones to work with.
With a 16 bit system, your first stop has 32,768 transitions to work with....a heck of a lot more than 128.
Considering how cheap storage is nowadays, why spend all the time scanning to capture a low quality original. The only way it could be worse is if the 8 bit scan was save as a quality 4 JPG.
Robin-T
Old skool newbie
Back to shooting and I find this very informative. So, You can scan 35mm then print from the scanned image....more info please.
Bob Michaels
nobody special
Please think about how a scanner actually functions
Please think about how a scanner actually functions
Has everyone realized that all scanners do the actual scan as RGB? Makes no difference if it is a b&w neg, color neg, color slide. No matter what settings you have, it simply shines a light through the film and reads the red shadows, the green shadows and blue shadows. Then this data is captured as an internal raw data file. And it does it the same way regardless of any settings.
Then this RGB internal raw scan data file, unimpacted by any adjustments you dial in, is sent back to the CPU where it is adjusted by the scanner driver, just another software application
The scanner driver software takes that raw scan data and manipulates it in the CPU so that it can be output as RGB or simply averages the channels and outputs as grayscale. That same process makes the output a positive or negative of what it read. It also makes the output file either 8 bit or 16 bit without regard to the precision of the actual scanner (sometimes 14 bit). This is the part that Vuescan, SilverFast, Nikon driver and Minolta driver all do slightly different.
Now if you are scanning a b&w neg (or even a b&w chrome) the red channel, the green channel and the blue channel look amazingly similar. Just as you would expect from converting gray into RGB. That is why the simple averaging of the channels into a grayscale file works so well.
While scanners are somewhat complex optical-electrical devices, they are not black magic with voodoo going on inside. It is important to know what the controls actually do and don't do. And what the scanner itself does and what the scanner driver software later does in the CPU with that data.
Personally I use Vuescan with my Minolta MultiPro and always have it output 16 bit grayscale positive files from b&w negs. I do almost no tweaking of the scan parameters. I only make sure the ends of the histogram are not clipped. I've tried all the various ways and that is simplest and works as well as anything else.
Please think about how a scanner actually functions
Has everyone realized that all scanners do the actual scan as RGB? Makes no difference if it is a b&w neg, color neg, color slide. No matter what settings you have, it simply shines a light through the film and reads the red shadows, the green shadows and blue shadows. Then this data is captured as an internal raw data file. And it does it the same way regardless of any settings.
Then this RGB internal raw scan data file, unimpacted by any adjustments you dial in, is sent back to the CPU where it is adjusted by the scanner driver, just another software application
The scanner driver software takes that raw scan data and manipulates it in the CPU so that it can be output as RGB or simply averages the channels and outputs as grayscale. That same process makes the output a positive or negative of what it read. It also makes the output file either 8 bit or 16 bit without regard to the precision of the actual scanner (sometimes 14 bit). This is the part that Vuescan, SilverFast, Nikon driver and Minolta driver all do slightly different.
Now if you are scanning a b&w neg (or even a b&w chrome) the red channel, the green channel and the blue channel look amazingly similar. Just as you would expect from converting gray into RGB. That is why the simple averaging of the channels into a grayscale file works so well.
While scanners are somewhat complex optical-electrical devices, they are not black magic with voodoo going on inside. It is important to know what the controls actually do and don't do. And what the scanner itself does and what the scanner driver software later does in the CPU with that data.
Personally I use Vuescan with my Minolta MultiPro and always have it output 16 bit grayscale positive files from b&w negs. I do almost no tweaking of the scan parameters. I only make sure the ends of the histogram are not clipped. I've tried all the various ways and that is simplest and works as well as anything else.
Last edited:
Spyderman
Well-known
A good idea might me to scan in RGB 16bit per channel and then choose only one channel and use it as 16bit greyscale. The point is that the red channel is usually very noisy, so when you desaturate, or make an average of R,G,B values, you are making the result noisier...
But I don't mind the noise, so I scan in grayscale.
But I don't mind the noise, so I scan in grayscale.
nrb
Nuno Borges
Now I only scan as color positive color negatives I developed in bw chemistry. For regular bw negatives I prefer to scan them as bw positives.
ampguy
Veteran
Why scan as a positive, then invert in PS? Is this a PS thing?
John Robertson
Well-known
Same info as I was given by a guy who lectured to us at our camera club a few years ago, I think he was called Barrie Thomas, he worked for one of the UK photo mags. He said it also gave better results for toning.
nrb
Nuno Borges
One reason only. It's faster to scan as positive all negatives and then invert in photoshop.