The Whole Gamut

0 Shares

I have been saving up for an Apple G5 for quite some time now. Yes, they are expensive but given how intuitive they are when it comes to using graphical programs like Adobe's PhotoShop, it's worth the price. Plus, my handy Powerbook (Apple's laptop) is starting to show signs of wear. It's a little over 4 years old now (in Internet terms, it's a grand-daddy).

I have been doing some research. True to form, every six months or so, Apple releases a new version of either a machine or an operating system. There was Jaguar, which is what I use. Then came Panther (Apple wanted $129 for it and I said “let's see … let's see”). To be released in early part of next year is “Tiger”. Well, at this rate, they'll run out of big cats. But back to my G5 purchase. It will happen this year and I am sure of it. Unless of course Apple releases a G5 laptop. Then it will be back to the drawing board to see the pluses and minuses of each configuration. What joy!

As a photographer, I need the RAM and the disk space. But I also need to edit images in 16-bit color. According to Kevin Yank of Sitepoint.com, Adobe's PhotoShop CS works wonders in that regard.

16-BIT COLOUR IN PHOTOSHOP CS

Photoshop CS is the first version of this venerable product to fully support 16-bit colour across its entire toolset (with the exception of some filters). But what exactly does that mean to Web developers, who continue to publish to an 8-bit medium?

A 24-bit image (e.g. a JPEG or PNG image) uses 8-bit RGB colour — 8 bits each for the amount of red, green and blue that goes into each pixel in the image — allowing for up to 16.7 million colours. Simply put, 16-bit colour means more bits per pixel, and more colors. 16-bit RGB colour produces 48-bit images, which allow for 281.5 trillion colours!

But in the arena of Web publishing, everything is displayed on a computer screen, where the operating system typically supports only 24 bits per pixel (i.e. 8-bit RGB colour). So what use is it?

The advantage of 16-bit colour lies in the ability to make significant changes to an image without losing as much image information as you would working in 8-bit colour.

Allow me to illustrate: Say you've got a digital photo and you want to apply a spotlight effect to it, cloaking the photo in darkness and revealing a part of it with a fuzzy, circular mask.

Now, experienced Photoshop users have an instinctive feel for how to go about an job like this so as to maintain the highest image quality. They'd start by selecting the area for the spotlight, then they'd invert that selection to cover the area outside the spotlight, and finally they'd use Image > Adjustments > Levels… to darken the selected area.

A more naive user might first darken the entire image, then select the area of the spotlight and attempt to brighten it to reveal the original image brightness. If the image uses 8-bit colour, the results from these steps will likely be disappointing. But 16-bit colour is another matter!

The best way to see the difference is by watching the image histogram (Window > Histogram).

The horizontal axis of the graph represents the range of colours in the image (specifically, their brightness), while the vertical axis represents the number of pixels in the image for each of those colours. A well-exposed digital photo will usually contain pixels spread across the entire range of brightness values.

When we darken the image, we compress that full range of colours into a more restricted range.

In an 8-bit colour image, pixels that were nearly the same colour in the original image will be forced to have the exact same colour in the darkened image. This translates to a loss of detail!

Brighten the image back up to its original levels, and the histogram looks different than it did to begin with. The outline of the graph is the same, but instead of a solid distribution, we see isolated vertical bars with gaps between them.

The detail that was lost in darkening the image (forcing a full spectrum of colour values into a smaller range) does not return when we brighten the image. Pixels that were forced to become the same colour remain the same colour in the brightened image, so we still have a reduced number colours. In extreme cases, this effect is easily visible as patches of solid colour where smooth tones previously existed.

But here's the punch line: If we simply covert the image to 16-bit colour (Image > Mode > 16 Bits/Channel) before making these changes, the final histogram looks much healthier. Indeed, the histogram (and the image itself) is almost indistinguishable from the original!

When we convert our image to 16-bit colour, we “make room” in the histogram for many more colours than the image actually has. In fact, for every one of the 16.7 million colours that can appear in an 8-bit color image, there are 16.7 million other colours that cannot appear in an 8-bit image. Think of this as adding 16.7 million empty spaces between each of the vertical lines that make up our original image's histogram.

When we compress the spectrum of colours in the image down to darken it, all those extra colours (i.e. all that extra space in the histogram) means that pixels that were almost the same colour in the original don't have to become the same colour to fit into the darkened image. All that detail is preserved, so when we brighten up the image again, we get the same detail and range of colour that we started with!

Of course, 16-bit colour has its downsides. The most significant is that, by definition, a 16-bit colour image will always be twice the size in memory as an 8-bit colour image of the same dimensions. Another is that many of Photoshop's filters have yet to support 16-bit colour.

But if you don't have the experience it takes to minimize lost detail as you work in Photoshop, or if you prefer to work without the pressure of doing things “in the correct order”, 16-bit colour can be a powerful tool.