RAW’ Let’s Get Dirty

In this weeks venture into RAW, today we will get into the theory of the digital capture process.

Digital Sensors - Before we can get into the processing of digital data, it helps to understand where why, and how that data is gathered. Below are two sensors with a relative size of a full frame and an APS-C (1.5 or 1.6 crop factor) sensor. The red, green and blue pattern is called a Bayer filter after a Kodak inventor by the name of Dr. Bryce E. Bayer. There are twice as many green filter elements as red and blue to match the human eyes resolving power with green light (Google angular resolution for more). Further due to the sizes of the sensors with the same number of pixels, the larger size sensor can sample a larger slice of light.

Lossless Data - This is rather simple. If you make a change to an image and can go back and make it the way it was, it is termed to be Lossless data. On the other hand if the image can not be reverted back to the original, data is lost and is termed Lossy data. This is what a JPEG file does as indicated by the diagram below.

Bit Depth - We will get real cyber-geeky for a second. First off one bite consists of eight bits therefore the binary code of 00000001 has the decimal equivalent of 1. Thus 00000010 = 2, 00000100 = 3 and so on. Combining all these various combinations of ones and zeros in one bite will come up with 256 combinations. It works by doubling the number with each bit (e.g. 2-4-8-16-32-64-128-256). It starts with two and must be an even number and zero does not count as it can not be a multiplier.

So what does this have to do with capturing a beautiful landscape, flower, or face? The camera sensor in most DSLR’s today captures 12 bits of data per channel. What these numbers represent are brightness levels or tonality. So 12 bits provides for 4096 tonal steps of data. The Canon 5D Mk II captures 14 bit which equates to 16384 tonal steps of data.

UPDATE 3/1/09

Here is a really good graphic of what I have explained above. Thanks to a post by Jim DiVitale at Digital Imaging Tune-Up Clinic. Read the whole post here.

A RAW image is further sampled to 16 bit which provides for 65,536 tonal steps per color channel. So when a JPEG is used, the computer looses 65,280 possibilities to adjust the image. Gone... Nada... Nothing… History Bubba!

Hopefully the graphic below shows what happen “In camera” when capturing to JPEG vs RAW.

Armed with this basic understanding as of capture, tomorrow we will get into some of the editing tools in Adobe’s Camera Raw.

C U Then… Doug