Digital Printing Insights #2: “How important are 16-bit scans over 8-bit scans?”

“How important are 16-bit scans over 8-bit scans?” This is a good question that I receive fairly regularly. First things first: An 8-bit file (8-bits per color channel [R/G/B]) contains only 256 data points per color channel. A 16-bit file contains up to 32,768 data points per color channel. Trick question: which bit mode contains more extensive and workable data? I knew you’d get it! I’m no mathematician myself, so to get right to the nitty gritty: when you’ve got the option and the money, always choose a 16-bit scan over 8-bit. You want MORE data to work with, not less. When one really starts “cranking” on their files (large brightness or contrast adjustments, for example) the additional bit depth becomes obvious.

16-bits

16-bits

Take a look at my 16-bit sample at left. I added one gross contrast adjustment via a curve adjustment layer in ‘Normal’ blending mode (when I say “gross” I mean it! The hyper-color saturation would be greatly controlled by switching to Luminosity blending mode, but my experiment would fail with this example). Notice that even after this adjustment, the histogram appears “uncombed” and good (“combing” = the choppy separations that occur between the histogram data). Now take a look at the 8-bit sample below it, also at left. I dragged the same curve adjustment layer from the 16-bit file to the 8-bit file. Take a look at this 8-bit histogram. See the combing? Combing of the histogram = lost data, and I know you don’t want this. You can try this experiment at home.

8-bits

8-bit

Photoshop now offers extensive support for 16-bit editing, so when cost, availability, and time is not an issue, always go for a 16-bit scan. What if you don’t have lots of disposable income to shed on 16-bit scans and you need to be a bit more selective? My general rule is this: if it’s a difficult negative or chrome (meaning under/overexposed or other such imperfections); if you have a fair bit of sky; or large areas of continuous tone (think water and reflections in water); or if you plan on “cranking on” the image in post-production (large contrast/levels adjustments), then 16-bit is definitely a better bet. With skies, water, and continuous tones, banding is always a possibility – especially at 8-bits and with large adjustments. Better data in = better data out.

I now scan everything 16-bit, but I still have a large number of files I print from that began as 8-bit scans. Some of them are just fine – no rescan necessary – and some would benefit by rescanning and reworking with 16-bits.

If you have any questions, comments, or suggestions for future Printing Insights, I would love to hear them! Thanks for reading.

You are visiting the blog of fine art photographer Michael E. Gordon. For additional photos and information, please visit his official website.

Advertisements

7 thoughts on “Digital Printing Insights #2: “How important are 16-bit scans over 8-bit scans?”

  1. Thanks for your comment, Ernie. I should have clarified further, but wanted to keep the article simple (I can’t with engineers like you hanging around! :)).

    My resource is Martin Evening, author “Adobe Photoshop (insert your version here) for Photographers”: “Photoshop’s 16-bit depth is 15 bit + 1.” At least with Photoshop, it uses only 32,768 of those 65,536 levels because this tonal range “is more than adequate to describe the data coming off any digital device.”

    Now you now where I got that figure from. Thanks for pointing this out, Ernie.

  2. Hey Michael,

    I seem to remember asking you this recently. 🙂

    I’ve been thinking more on this as I’m still trying decide on a new scanner (either drum, eversmart, or cezanne).

    I didn’t realize the 15bit assumption made by photoshop. As far as scans go, I know of no analog to digital converters (ADCs) that are bigger than 14bit. We’ll see what that means for future digicams, but as of now, I think Adobe is safe with their assumption.

  3. Hi,

    If I get 2 scans from the same TX, a 16 bit and 8 bit drum scan, and then convert the 8bit scan to 16 bit in CS3, will I get the same information in the file as with the 16 bit scan?

    If not, is there any value in pumping up the 8 bit scan in CS3?

    Thanks,

    Sam

  4. Hi Sam: converting an 8-bit scan to 16-bit gains you nothing but an inflated file size, as those additional 8 bits must be interpolated from the original 8-bit scan. The result may be less sharpness, less detail, and the potential for artifacts.

    In general, it’s best to always scan at 16-bit depth and work with this as your master file. One should only convert A COPY to 8-bit if your output dictates it or if your computing speed is not great enough for the file size (keep the master file at 16-bit!).

    I hope this helps!

Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out / Change )

Twitter picture

You are commenting using your Twitter account. Log Out / Change )

Facebook photo

You are commenting using your Facebook account. Log Out / Change )

Google+ photo

You are commenting using your Google+ account. Log Out / Change )

Connecting to %s