Out of Gamut: Why Do High-Bit Editing?
|
Almost everyone who has a scanner does some high-bit editing whether they know it or not. Just about every scanner being made today captures at least 10 bits per channel, and when you use the scanner’s software to make edits, it’s operating on that high-bit data. It’s important to make your big tone and color moves on the high-bit data, because if you don’t, you’re wasting the extra bit depth you paid for with the scanner, and you’re more likely to encounter posterization or color banding. Why?
Anytime you edit a file that has 8 bits, and hence 256 levels per channel, you wind up with fewer levels in each channel than you started with. When you move the midtone slider in Photoshop’s Levels dialog box from 1.0 to 1.2, it remaps level 128 to a higher value (144 in a gamma 2.2 working space). So the levels from 0 to 128 get stretched to encompass the new range of 0 to 144. In the process, some levels wind up with no pixels (in this case, levels 1, 3, 5, 7, 9, 12, 15, 18, 22, 26, 30, 35, 41, 47, 55, 64, and 79 all end up with no pixels). And the levels from 128 to 256 become compressed, so that the 127 levels from 128 to 255 turn into 111, because some of the pixel values that were formerly different get squashed together. All told, you’ve turned 256 levels into 223.
This isn’t necessarily a problem, but the holes you’ve opened up in the tonal range will be further exacerbated by future edits or color space conversions, and the levels that got squashed together represent difference, and hence potential detail, that’s been lost irrevocably.
When you edit a high-bit file, however, each channel contains 65,536 potential levels, even though only the 256 most significant ones get displayed. Edits still lose levels, but you have so many more to play with that the loss is much more affordable, since you still wind up with a full 256 shades per channel.
One way to accomplish this is to make your major edits in the scanner software. But most scanner drivers give you a very small preview window on which to base your decisions, and their tools are often much less sophisticated than Photoshop’s. So another way to accomplish this is to bring the high-bit data into Photoshop, and edit it there, where you can see all the pixels and use Photoshop’s tools instead of the scanner driver’s.
At some point, you’ll need to downsample the image to 8 bits per channel (even if it’s not until Photoshop sends the data from the high-bit file to the printer driver), but if you do the bulk of your editing in high-bit mode, you’ll end up with a “perfect” 8-bit/channel file that uses all 256 levels instead of one that may contain only 180 or so.
It’s up to you how far you take your images in high-bit form. If you need to make edits using blending modes, or do serious compositing, you’ll need to do these in 8-bit mode, but you’ll run into fewer problems with an ideal 8-bit/channel image than with one that’s already suffered significant degradation. But if you’ve been downsampling because Photoshop didn’t appear to let you do something in high-bit mode, the techniques in the main bar may lead you to the conclusion that you’ve been downsampling before you really needed to.
 |