Intro to Image Sensor Capture:...

Intro to Image Sensor Capture: ENLIGHTEN US!
One standard of raw data? Think about the varying grid patterns of image sensors:
Most use 4-pixel blocks with 2 green, 1 red and 1 blue in the form of CCD or CMOS.
Fuji has that plus between each pixel is a smaller, less sensitive pixel with their SuperCCD SR sensor.
Sony's got a funky four color CCD where the second green is "emerald" instead.
Sigma uses something more film-like with the Foveon X3, where the pixels are stacked directly Blue on top of Green on top of Red.
My point being: Don't even RAW images get interpolated when saved in-camera? And if not, could one standard format of raw data be established? And why DNG?

James Maher – Mon, 2005/04/25 – 4:43pm

Raw formats AIUT are a direct lzw compressed data dump of...

Raw formats AIUT are a direct lzw compressed data dump of the CCD, which has to be interpreted back to a DIB for display.

The Raw data is not interpolated until it get to the decoding engine, be that in camera or in the PC/MAC.

There should be an additional header in the Openraw format Specifying the Bit Pack Structure

IE

RGGBA: RED GREEN GREEN BLUE ALPHA

or The specific arrangement, hence making it easy for the developer to interpret the image.

Jacob AA Alifrangis – Mon, 2005/04/25 – 5:02pm

The formats that DNG supports are the same as those of...

The formats that DNG supports are the same as those of Camera Raw, which are at:

http://www.adobe.com/products/photoshop/cameraraw.html

They include cameras from Fujifilm, Sigma, and Sony, as well as the usual suspects. Do these include the ones you mention? DNG has no trouble supporting a variety of sensor formats. And DNG can support de-mosaiced (interpolated) sensor data too.

"Why DNG?".

What is important is that having a controlled documented standard for Raw is part of the solution to the goals of this website. DNG may be that standard, or a step on the way to that standard, or a proof of concept that such a standard can exist.

Barry Pearson – Mon, 2005/04/25 – 6:35pm