Interpolation?

[dave52240]

Hi all, I have recently bought a 6 mp digital camera, the resolution on the high setting is quoted as 2976 x 2232 {by interpolation} what is meant by interpolation.

[another view]

Interpolation is the same thing as up-sizing. If it's done right and not up-sized too far, it's hard to tell that anything was done. This is handy for making big prints. But with those numbers I get 6.6 megapixels - it must be a lower-resolution camera that interpolates up to 6.6mp?

[Peter_AUS]

Dave it means that a calculation is done on the surrounding pixels and creates non existent pixels to fill in the image to create the 6mp final image, so if the camera takes the image say at 3mp, the programming inside the camera then calculates using each sourounding pixel to insert other pixels based on it's best guess, based on thousands of calculations from the companies past known information. This has been done for a long time and there are programs around that do it quite successfully. Photoshop can do it, if you increase the size by 10% each time you increase the size, using the options to interpolate it uses the surrounding pixels and fills in the extra 10%.

Hope that makes sense.

[dave52240]

Many thanks to you both for the clear explanation,and yes, it actually is 6.6 mp.

[Skyman]

the classic problem with interpolation is that the layout of a conventional ccd cannot average pixels effectively and still maintain natural looking images. a regular ccd looks something like a chess board. (square pixels stacked next to each other in a grid) when a camera tries to interpolate the image off this sensor, the algorythms have problems blending the colours (an area of blue next to an area of yellow will often interpolate to give a fine area of green between the two) the fuji cameras with the super ccd (fuji's brand name) use hexagonal pixels and interlock them like a brick wall to give superior performance when interpolating images. fuji claim that the process they utlise is not interpolation and in a way this is true, the sensor actually analyses the rate of change in the wavelength of light across the pixels to predict the wavelengths of light in areas of the sensor where there are no pixels. not really interpolation as the math is different but the same end result.