Originally posted by mysterick Can you further elaborate please?

DR is normally defined as the number of luminance increments between white and a gray level with 50% (or some other fixed ratio) noise.

To start with, JPG only has 256 luminance increments at all (8 Bit). Without strong shadow compensation, this will have already limited the DR to ~8 EV.

The K-7 offers optional strong shadow compensation and this may be of lesser concern when used.

Now, let's consider RAW files.

Say, a RAW file has 12 Bits or 4096 luminance increments.

The darkest gray would be 1, white 4095.

If you

*properly* scale the image to 50% size, the darkest possible gray becomes 0.25 (from 3 black and 1 dark gray pixels). Theoretically, you get 16384 luminance increments in the smaller image. Or two full stops more DR.

However, RAW files are such that the darkest grays always are totally noisy (random) rendering the above consideration completely irrelevant

What is relevant though is this: Due to averaging, the noise of the average of 4 combined pixels is only 50% of the noise of a single pixel (Poisson law).

Therefore, you gain one full stop of DR when sizing to 50%. Therefore, a web-sized image (600x400) has three full stops more DR than the original sized image (requires 16 Bit TIFF of course).

So, even if the sensor outresolves the lens, you still shouldn't shoot smaller JPGs because this destroys dynamic range.