16 bit vs. 8 bit when scanning - is it worth the hassle?

[photom]

Is scanning at 16 bit vs. 8 bit really worth the hassle? I just

started doing scans of 6x7 film on a Nikon 9000 (4000 dpi) and I

notice that it really slows things down when at 16 vs. 8 bit. I have

read that there are some advantages when doing adjustments in

Photoshop by using 16 bit as to preserving the image. I cannot image

doing a 4000 dpi scan with ICE, 16 bit, and 16x multi-sampling (would

take a long time). I realize that 1 gig of RAM is not much anymore -

I guess I have an upgrade to look forward to next year. Curious what

others have found regarding this.

[andrew_rodney1]

It's worth it if you're doing global tone and color tweaking and don't want to lose any

information due to the edits! Do as much as you can in the scan software which should

apply the edits on the fly to high bit and save out 8 bit to speed things up. There are

however better tools available in Photoshop so you might want to do this in high bit from

the scanner and suffer the speed loss.

[michael_spencer3]

Tom, using a 16 bit file doesn't really "preserve the image". It really is meant to reduce the amount of posterization that might result from a color or contrast adjustment. If you do the adjustments at 12 or or 14 bits and THEN change the file to 8 bit you will get a smoother histogram. Of course, you have to get it to 8 bits in order to get it printed. Whether this clumping due to the color/contrast adjustments would have any visible affect on the actual appearance of the printed image is another issue all together and would depend on the particular image and the amount of "clumping" actually created.

 

I scan 35mm slides mostly, so in my case that's only a 100mb file. I find it is really no problem at all to scan (with color/contrast adjustments) in the 48bit color mode and then save as the 24bit TIFF. Nikonscan does this conversion and file save much, much faster than doing this in PS; I try to keep the 48bit files away from PS.

[walterh]

Tom only you can decide if its worth it. If you see no difference in the result its a rather simple decision. If you see a difference like I do in most scans of negatives then only you can define how much the "little" difference is worth to you in time.

 

If your question was : do I se a difference in quality? Yes I do. In simple terms - the more dynamic range within the negative and the better the quality the more it matters. With some practise you will get a feel which negative will be longing for the extra resolution. I usually prescan all images fast with low pixel resolution and 8bits . Then I decide wich ones are worth the extra effort. By the way - for some images it may be even "worth" it to try and scan using different software for scanning (Silverfast/Nikon driver or software/vuescan). But this is only to get the best out of it.

Cheers

walter

[mag_miksch]

You see the difference between RGB and sRGB where?

[photom]

Thanks for the valuable info. I think I will try to do as many adustments in Nikon Scan as can be done, if I think I can get away with it, and then save to 8 bit. Not sure yet how well the shadow/highlight adjustment and USM work at the Nikon Scan level vs. PS.

 

Mag - not sure what you were asking. Might be another discussion topic?

[qtluong]

My experience has been that if you don't scan at the maximum quality possible with your scanner (which in your case is 4000dpi, 16bit, 16x) at some point you will rescan. What's the problem if it takes a long time ? It's the computer that is working, not you.

<a href = "http://www.terragalleria.com/">Terra Galleria stock photography</a>

[michael_chmilar]

It really comes down to: Where do you prefer to work? In Photoshop, or in the scanning software?

 

Initially, with any scan, the first thing you do is set your "exposure", ie. where is the black level, white level, and midpoint. It is essential to do this with 16-bit channels; otherwise, you will get a "picket fence" histogram, which means that you have lost some information which you cannot recover. (The importance of the lost information depends a lot on the content of the image.) The "picket fencing" will only get worse as you perform subsequent adjustments.

 

After you have dialed in the initial levels, most of the adjustments you make in Photoshop can be done reasonably with 8-bit channels. There are exceptions, which may cause problems in 8-bit mode: extreme adjustments; long, smooth gradients; etc.

 

So, you can either:

 

Get your levels set in the scanning software, and save 8-bit files for Photoshop.

 

or

 

Don't use the scanning software for any settings. Just save out the 16-bit without making adjustments. Load the 16-bit file into Photoshop, and set your levels. At this point you can either convert to 8-bit and continue working, or, if you have enough RAM and horsepower, keep working on the 16-bit data.

[bobmichaels]

Your printer is going to convert to an 8 bit file before it prints so it's entirely a function of how major adjustments you do to the contrast and levels in PS. Typical adjustments to a properly exposed, developed and scanned neg seem to suffer no loss when scanned at 8 bit to my eye.

 

But, I would ask why you are scanning a 6x7 neg at 4000 dpi? How big are you planning on printing? Remember your printer will downscale to 360 dpi so do the math (or cheat and use PS as a crutch) Your 4000dpi scan will get you something like a 20x24 print at 360 dpi. And a very large print does fine at much less dpi.

 

Some experimentation with critical viewing of your actual prints may show you that you've got a lot of overkill in all the factors that cause huge files. Now, having said that I still sometime find myself scanning 6x6 negs at 16 bit 4200 dpi "just in case". But I find myself immediately converting the huge scan TIF file to something more manageable when I open it in PS and I have 1.5 Gb of RAM.

[link]

Tom,<p>

 

I have the 8000 model and I scan at 4000dpi/16bit/Ice/fine mode.<p>

 

I don't scan slides often and multisampling doesn't do anything visible when scanning

negs. For slides without a lot of dark areas, multi sampling is probably not

needed.<p>

 

Sure, my scans take a bit of time, but I like to do all the color/density corrections in

photoshop. My 6x9 scans make a file of about 650mb. <p>

 

To make photoshop workable with such large files (and they get much larger when I

work on them!) I do all my adjustments as adjustment layers. This leaves all the

rendering of the large file to the last step before printing. Since I use Photoshop CS, I

now keep my master files in 16bit mode, with all the adjustment layers saved. For

perspective, I can now save 2 photographs on a DVD.<p>

 

Why work with such large files?<p>

 

I don't print that large - yet, but I put a lot of work into the master files and would

never want to rescan and duplicate all that work to make a large print. I also revisit

the files sometimes and like to make changes. Keeping all the adjustment layers

allows me to save all the original data from the scan.

 

I do have 2gig of memory and when photoshop can handle more, I'll upgrade to a

computer that can handle the additional memory (mac G5).<p>

 

The price I pay is that the scans can take about 20 minutes, and saving the master

files with adjustment layers can take an hour sometimes. Rendering for printing can

take a while too...<p>

 

-bruce<p>

<a href="http://www.brucealangreene.com/photographs.html">I have some

examples on this web page</a>

[lachaine]

I didn't find it made any difference, personally, but I only did minor editing of the file. If it's for web display, it all ends up looking the same anyway. If it's for printing, then, your best bet is to experiment for yourself.

[jan_brittenson]

I always make all adjustments in 16 bit. Once I have the final file I save it as an 8-bit TIFF. Anything printed is downsized, and it doesn't matter if this is done in 8 bits. 8 bits is enough even if it might lose half a bit or so in the output profile conversion depending on paper.

 

The most important aspect though is probably to get the black and whitepoints right at scan time. In film scanners this is done in analog hardware using programmable gain and offset in the ADCs. Since each channel has its own ADC it also has its own levels (gain, offset). This is highly preferable to having this adjustment made in software.

 

Going from linear to the final image gamma (e.g. 2.0) needs to be done after black and whitepoint adjustments, but before 8 bit sample reduction. Scan to 8 bit only if you are positive your gamma is real close to final. Avoid gamma-on-gamma changes; here, color managed software that can give an accurate rendition of hue (white, blackpoint) and tonality (gamma) is invaluable. As is scanning software that can scan to a raw file (linear) and allow you to go back and redo things like gamma without having to actually rescan. (e.g. Imacon .3F files.) Fortunately printers are more forgiving than CRTs, but converting to a print output from a reference CRT space (e.g. AdobeRGB) still changes gamma, and being able to go straight from raw-to-print with a CRT only for output soft proof (icc simulation) can be the difference between dark tone banding and no dark area banding in print. If you can't seem to get good shadows for a problem original no matter what, then scanning straight to the print space, bypassing the reference space, is often worth trying.