I don’t know if you can use the term "dramatically" in the amount of difference. Many people can see a difference (I don’t). Of course, many filters won’t work with 16 bit, and the files will be 4x as large.
I don’t know about converting, but bringing in original 16 bit images will give you better results for a number of things esp. having to do with color adjustments. I’ll let more knowlegable folks speak about converting from 8 to 16 for work on what would normally be an 8 bit image.
No, it will not. Just makes files larger.
Capturing the image in 16 bit is when you should consider this.
The advantage is for when you plan to do quite a bit of editing on the image and/or converting the file to many device spaces.
It’s really good for fixing bad exposures where you have to make large moves in the highlites and shadow regions. You’ll find you can make finer, smoother adjusts with curves tool, for instance, in very pinpoint spots without abrupt tonal transitions and exaggerated noise.
I use it for restoring photos that are destined for multiple edits.
Tim’s spot on, of course, but I wanted to emphasize that converting to 16-bit won’t help your images. If you bring them in as 16-bit (via camera or scanner for example), that’s a different conversation.
Peace,
Tony
Holy moly. That reminded me of when I watched "A brief history of time". I got that there is a universe, and that some guys disagree about whether there was a single point event or whether it always has been in existence. But I couldn’t follow the logic.
I’m not smart enough to decide based on that Harv. Really. I can’t even spell CPU. <shrug>
Sixteen bits are NOT provided to give even more tones (65,536) than the already-superfluous 256 tones generated by eight bits. The eye can’t even distinguish those 256 tones, let alone 65,536.
That is not to say that 16 bits are useless. But they don’t do anything for you UNTIL YOU ALTER the input-output curve when editing an image. Any deviation from a 45-degree slope of that curve results in a loss of tones in the zone concerned—the greater the deviation the greater the loss. The more tonal values you have to start with, the more losses you can afford before cutting into the underlying basic tonal data measured. In other words, sixteen bits helps to PRESERVE the tones that count when you massage the data by diddling with gamma and otherwise monkeying around with curves.
George
yes. what george said. that’s what i was getting at. 🙂
Likewise.
Only contention is the bit about "the already superfluous 256 tones". I don’t care what the physicists say, though the eye may not be able to distinguish the border between each of those 256 tones when presented together, a traditional (silver/gelatine) B&W photographic print with virtually infinite tonal gradation "looks" a hell of a lot better than an 8-bit (256 tones) digital print.
So the eye can easily distinguish the difference in an overall sense. The greater the bit depth (particularly when output technologies catch up), the better the print will look, though this is not all that relevant with present technology.
But George is spot on with the advice he gives.
Remember that the 256 tones are for each of the primary colors … ie, Red green and blue or cyan, magenta, yellow and black. Thus the number of colors in 8-bit is actually in the millions.
But not in greyscale: that’s 256, and it’s woefully inadequate.
Extrapolate that into colour, and the same applies. Whilst maybe not quite as obvious as in greyscale, the inadequacy is still there, as will be proved when practical 16-bit physical output is realized. It will be visually better.
George’s advice is sound from the viewpoint of starting with the maximum information before messing with it. In 8-bit one is hamstrung with barely adequate information to begin with, so editing it can result in unacceptable loss of information in the final result. 16-bit gives more leeway.
Without being intentionally pedantic: black is not a colour – it is the absence of colour, just as white is the presence of all colours. When referring to light (as opposed to paint) primary colours are red, green and blue, secondaries (complimentaries) are cyan, magenta and yellow. Black is a printer’s addition to make up for the fact that C+M+Y inks together produce a brown, not the black they are logically supposed to.
Fred beat me to it. If you think you cannot see one bit, open curves, pick a point say 128 for input, set 129 or 127 for output, then click the Preview box off and on.
I am working on a scan of a beach photo taken with an 8×10 years ago, and the beach tonality is very flat. Trying to obtain the subtle modulations of gradation in the sand that I clearly see in the silver print is a real headache. If it wasn’t for the damn light leak in a corner, I would not continue, and this image would remain a contact print on a silver based emulsion.
I told you this was a different discussion <grin>.
