On Mar 12, 2005, at 6:55 AM, Shel Belinkoff wrote:
I just read this statement: "Raw formats ... allow you to retrieve images
before any in-camera processing has been performed."
OK, that seems to be the general consensus, but isn't the original image
captured in B&W and then converted to color, and then moved along for
further processing? If that's the case, then isn't the RAW image and
information is already manipulated.
No, not exactly.
The sensor is a two dimensional array of photo sensitive sites that count photons. In front of each photosite is a filter, so each photosite is only going to see one of R,G or B light. The filter pattern for a sensor has twice as many green photosites as red or blue. What the sensor data for an exposure contains is a 2D array of numbers representing the photon counted at each of those photosites.
That array of numbers doesn't represent a "B&W" or color image in RGB space without processing. Two basic kinds of processing have to happen to obtain an image:
- Digital sensors see light in a linear gamma form... each photosite simply counts photons hitting it and produces a number in a 12bit range (0-4095). Our eyes, receiving the same data, compress the highlights and expand the shadow values. This is described by a gamma curve. So one thing that a RAW processor has to do is convert the linear gamma recorded by the sensor into the normal gamma curve that our eye would see.
- Again, the data from each photosite is an intensity level recorded in Red, Green, or Blue light. To convert this to a composite RGB or B&W image requires a Bayer matrix calculation be performed in order to evaluate by interpolation the correct RGB values for each pixel as represented by a photosite.
There are also several secondary transformations ... evaluating the white balance, contrast, saturation and sharpening constants applied in the Bayer and gamma correction operations ... but for simplicity's sake let's leave it to be these two primary operations.
These two operations together are what the camera does internally when you save exposures as JPEG or TIFF. When you save in RAW format files, the only processing applied to the sensor data is to write it in a recognizable format as an array and then collect up the camera settings used for the exposure (ISO, aperture, shutter, time, date, etc etc) and add them to the sensor data. Optional processing that can be done on the RAW format file is compression (stripping unused bits, packing and lossless compression operations to reduce the file size), but otherwise there is no transformation to RGB space: the full sensor data and camera settings can be losslessly recreated from the RAW file outside the camera for conversion to an RGB or B&W image in post-processing.
Godfrey
