Quote:
Originally Posted by Camelopardalis
Traditionally, large photon buckets (pixels) were necessary to achieve acceptable SNR from the stream of photons hitting the sensor, largely because of high read noise and significant dark noise.
The only problem with tradition is that over time it gets outgunned by the technological revolution.
CMOS sensors are now starting to get competitive with their forebears with a combination of high QE and low noise, resulting in (commonly) smaller pixels that are able to achieve similarly acceptable SNR from fewer photons. That doesn't mean that chasing the faintest of fuzzies doesn't take a long time, but what it does it mean is that it isn't impossible with more humble kit.
|
The main improvement has been a reduction in read noise (there are CCDs with high QE as well.) This allows the use of much shorter subs. It doesn't make a lot of difference to the total integration time required as you still have to overcome shot noise. Going large on aperture and pixels is still a winning combination for those who can afford it.
Cheers,
Rick.