Mar
12
Higher-Bit Images Don’t Increase Dynamic Range
Filed Under Color Basics, Digital Imaging, Herbiology
By now we all know that 8-bit images contain 256 levels of color spanning from solid to no color at all. And if we do the math (2×2x2×2x2×2, etcetera, ad nauseum) we see that higher bit-depths deliver exponentially-higher amounts of tone between solid color and no color. 9-bit images contain 512 levels, 10-bit contain 1024, and so forth. All those teeny-weeny steps just get smaller and smaller the higher up the bit-ladder we travel. All well and good.
But do you realize that for all that frothing about levels of tone absolutely nothing changes concerning the bookend colors themselves? Let me state this a little differently. Let’s start with an 8-bit RGB image that doesn’t have a great range at all. Perhaps it’s darkest color is R25G57B31 and its lightest color is R230G228B211. The dynamic range of this image is relatively low. In grayscale values that’s about 88% dark to 13% light.
Guess what happens to that shallow range when we move from 8-bits per channel mode all the way up to 16-bits per channel? Absolutely nothing! We’re still stuck with a contrast range that delivers all the excitement of a bowl of oatmeal. WOW! Think of the quality edge we just purchased with all that hard disk real estate.
OK, enough with the sarcastic talk. But do consider this…
The visual dynamic range limitations of any printing device (combined black ink density versus paper white) determine the appearance of any image sent through it. Here’s a fact to consider. The dynamic range of any printer is less than that of a glossy photographic print… 1.5:1. And simple truth is that little more than 200 levels of tone can be discerned by the human eye. Do some checking into this before you spend oodles of time adjusting for minute detail in high-bit mode.
Think about it!
©copyright 2008 Herb Paynter
http://www.imageprep.net
Comments
3 Responses to “Higher-Bit Images Don’t Increase Dynamic Range”
Leave a Reply
You must be logged in to post a comment.
When it comes to the Canon iPF ink jet printers… there is a dynamic range benefit in printing with the 16 bit option. But only if you have originally processed and maintained a RAW file in the 16 bit data mode.
It’s not how many bits you’ve got that equals increased dynamic range - it’s what they allow you to do.
If the notional image you mention is captured with a high bit-depth, you can stretch the limited range of shade actually contained in the image to fill the full 0%-100% (i.e. pump up the contrast) without it turning into a horrible posterized mess. And *then* you can send it to the (possibly crummy) printer. If you do this with an 8-bit image, the results are guaranteed to inferior to what your screen or printer is capable of reproducing.
One significant advantage of devices which capture high bit-depth images is that they are free to err on the side of under-exposing an image, guaranteeing that information is captured in the highlights, while knowing that significant detail will still be captured in the dark and mid-tones (all of which can then be worked with as the user sees fit - i.e. tuned to suit their display hardware, and the effect they are trying to achieve).
8-bit devices are forced to over-expose, causing the bright areas of the image to clip and be lost forever (like the image of the Sun, above), in order that the image contains any useful level of information in the lower ranges. Alternatively, they can sacrifice the lower range infromation in the name of doing the bright parts justice. But they can’t have their cake and eat it, because 8 bits is not sufficient to represent the range of intensities which tend to exist in the real world, and still capture the differences present in subtle textures.
That IS an effectively limited dynamic range.
All very true. My point was that a deep-bit capture in-and-of-itself does not produce a higher dynamic range image. There is a misconception with some that either capturing or editing in high-bit mode will mystically and automatically increase the dynamics of the original image.