>> |
Anonymous File :-(, x)
Protip: This is an image from one of Sony's finest scientific image sensors in my own "tinker-cam". One second exposure, low analog gain, sensor temperature about 40°C. I have converted the raw sensor data using only a nearest neighbor color interpolation and by linear mapping the 12 bit dynamic range to 8 bit. Images don't get much more raw than this. Notice anything? Camera manufacturers do not care if a few hundred (!) of all the mega pixels are defective. They just take an image with the lens cap on, store the position of every pixel that differs from the rest in a built in flash memory and from then on interpolate their value from its neighbors. Easy as pie, and you wouldn't notice anything, if it weren't for some pixels that decide to go bad after some time. Some cameras have an option to repeat the calibration hidden somewhere in their menus.
|