New
Year's Resolutions
By Mike Pasini, EditorImaging Resource Newsletter
Are there a few things you always meant to get straight but somehow they keep slipping out of your grasp? Well, we're going to resolve them for you before the millennium is out. We're talking about resolution, bit depth, and color space. Stuff you have to know to get your license.
Resolution
The following Resolution Comparison chart shows the difference between (approximate) uncompressed file sizes of several popular imaging resolutions, each presented in three different image modes: grayscale, RGB, and CMYK. (Note: each camera lists two or three resolution settings.) The value of the chart is not so much in showing precise file sizes, but in demonstrating how 2-megapixel CCDs compare to 1-megapixel and VGA CCDs (as well as film and Photo CD). Remember 1K is equal to 1,024 bytes.
| RESOLUTION | IMAGE SIZE (bytes) | |||
|---|---|---|---|---|
| CAMERA | CCD SIZE | Grayscale | RGB | CMYK |
| 35mm film | [grain] | - | 100,000,000 | - |
| PhotoCD | 3072 x 2048 | 6,291,465 | 18,874,368 | 25,165,824 |
| 2.1-megapixel Nikon 950 & Olympus C-2000 |
1600 x 1200 | 1,920,000 | 5,760,000 | 7,680,000 |
| 1024 x 768 | 786,432 | 2,359,296 | 3,145,728 | |
| 1.5-megapixel Kodak DC265 | 1530 x 1024 | 1,566,720 | 4,700,160 | 6,266,880 |
| 1152 x 768 | 884,736 | 2,654,208 | 3,538,944 | |
| 768 x 512 | 393,216 | 1,179,648 | 1,572,864 | |
| VGA | 640 x 480 | 307,200 | 921,600 | 1,228,800 |
If your camera isn't on our list, just use this formula to calculate the values
for your CCD: Multiply the CCD's pixel height by its pixel width (h x w) to
get a grayscale value. Multiple the grayscale value by three (red, green, and
blue) to get the RGB value. Add the grayscale value to the RGB value (one more
channel) to get the CMYK value.
If your browser has Javascript enabled, try the Image Size calculator below (and Merry Christmas).
Bit Depth
We're switching to monitors now. Not just monitors but system video, which can be provided on the motherboard or by an add-on card (usually offering some sort of video acceleration).
You may never know how good your camera is if you're viewing your images on a system that can't display all the color information captured by the camera.
Your camera records 24-bit RGB images. Which is to say that for every pixel or picture element of the image (every spot), your camera will record a value for red, green, and blue light that can vary in intensity by up to 8 bits. Three eights (3 x 8) are 24, hence 24-bit. So if the camera records 4 bits of red, 1 bit of green, and 7 bits of blue for one spot, the resulting color is purple.
This RGB data drives the red, green, and blue guns on your CRT like dimmer switches. Turning on just one bit out of eight is pretty romantic, while turning them all on is searchlight stuff.
Any color monitor is capable of displaying a 24-bit RGB image -- but not every
computer is. Some computers have only enough video memory to display a 16-bit
image (5 bits for each gun), some monitors display only 8 bits, for a total
of only 256 colors. Here's how it breaks out:
| BITS | COLORS | NOTES |
|---|---|---|
| 24 | 16,777,216 | aka millions of colors, true or full color |
| 16 | 65,536 | aka thousands of colors |
| 8 | 256 | where color imagining starts |
| 4 | 16 | when the Mac was B&W and the PC had color |
| 2 | 4 | black, white, plus two grays |
| 1 | 2 | black and white |
If you don't have 24-bit color on your system, you may be able to:
- Use a lower screen resolution so the video RAM can be used more generously for color. With just 1MB of video RAM, you can still get 24-bit color if you use a 640 x 480-pixel screen. With 2MB, you can use a finer 800 x 600-pixel screen display, and with 4MB, you can have a 1024 x 768-pixel display.
- Install more video RAM.
- Install a video card that supports 24-bit color
But you may not be able to push your video system to 24-bit. It depends on your computer model. Don't despair. Quite a lot can be done in 8-bit with intelligent use of the color palette. Some years ago the paintings of the National Gallery in London were digitized for a popular CD, using nothing more than 8-bit color.
And what about 32-bit color? Well, behind the RGB curtain, you'll see it's nothing more than 24-bit color with an 8-bit alpha channel, perhaps used for transparency.
Color Space
RGB and CMYK are the two most common color spaces. Got a color television? You've got RGB. Got a color magazine? You've got CMYK.
What's the difference?
Well, the television tube transmits color and the magazine page reflects it. Transmitted color is handy in a dark room, you can watch television. Reflected color isn't quite as useful in a dark room, because all you'll see is black.
Your camera records RGB color information. But your printer has to print CMYK (After all, it's making pictures without batteries, so the light is going to have to be borrowed, or reflected).
And this causes problems.
Partly because the RGB color gamut (or range of colors) is larger than the CMYK color gamut. Which is to say, simply, that you get a lot less colors in CMYK than you do in RGB. Your printer can't mix its four inks to match what that dimmer switch of a monitor can display. Which is why the pictures you print don't always match the image on your screen.
Unfortunately red-eye is well within both color gamuts, so it will come out just fine.
This article is reprinted from The Imaging Resource Digital Photography Newsletter,
Advanced Mode Column, published December 17, 1999