ejmartin Offline Upload & Sell: Off
|
Without seeing the raw data, there are several possible sources for the noise. One is overall underexposure, low S/N due to low signal which will make all color channels noisy as exposure compensation is applied during raw conversion -- amplifying the signal to bring it to proper exposure amplifies the noise along with it. Another issue is white balance; the red channel can be a stop or more underexposed relative to the green channel in the raw data due to the spectral content of the light illuminating the scene. White balance during raw conversion (either in the camera's jpeg engine or in a raw converter) again amplifies the red channel data to achieve perceptual white balance, and again amplifies the noise along with it.
LCollector: The spatial frequency dependence of noise in any color channel is an artifact of the raw conversion process, specifically demosaicing of the Bayer data. The spatial frequency dependence of noise in the raw data is uniform ("white noise") in the raw data, since the noise is uncorrelated from photosite to photosite (to a good approximation for read noise, and exactly for photon shot noise). The folks at IR and DPR would do well to understand what their tests are actually measuring. Often it's properties of the raw converter or jpeg engine and not the capabilities of the camera design.
|