Originally posted by Class A Where else would it be generated?
I already answered this by giving some examples.
Originally posted by Class A The sensor delivers digital data. There is no way any noise could be introduced outside the sensor.
Whoa there. The digital image capture process is largely analogue (one could say the film is digital by the way) - that is, until the A/DC does its job. Only down the stream it is digital.
The sensors are so incredibly efficient than any itsy bitsy factor can introduce significant noise in this chain. I already gave some examples.
Originally posted by Class A Of course, there is the noise of the light itself, but I think we are both excluding that here.
Of course.
Originally posted by Class A Of course there are external factors that influence the noise, for instance temperature.
These have nothing to do with the "accelerator" unit, though.
Do you even know what the accelerator does? What if it does apply dark frame NR based on - for instance - temperature, or other such factors?
Originally posted by Class A Every manufacturer knows how to build a power supply that meets the requirements of a Sony sensor.
Whoa there #2. It's much more complicated than writing the line above.
Originally posted by Class A And, again, what does this have to do with the "accelerator" unit?
Whether anyone makes certain assumptions about the power supply being able to induce noise or not, has no bearing on what the "accelerator" unit does.
Again, do you even know what the accelerator does?
Originally posted by Class A See above. Not relevant for what the true function of the "accelerator" unit is.
Which is...?
Originally posted by Class A If the "accelerator" unit were able to mitigate any such influences then you'd be able to achieve the same with a K-1 by just stabilising the operating conditions.
Again this is nowhere as simple as writing the line above. "Stabilising the operating conditions"? What the beep does it means?
Originally posted by Class A This patent describes an approach in which the "noise-reading processor reads noise signals from a pixel circuit for a pixel that is an object of noise acquisition, in parallel with the reading of accumulated charges by the charge-reading processor.". This is a good idea to deal with transients during sensor readout but it is not possible to implement this scheme with a stock Sony sensor.
The patent was an example of noise issues actually encountered in a camera, because you have this wrong idea that a sensor is impervious to external (internal to the camera) factors.
It might be applicable to a Sony sensor, it might be applicable to a Ricoh custom-made sensor. I won't try to overload my brain with automatically translated Patentese to find out which is more likely.
Originally posted by Class A No one is disputing that you prefer the processed images.
Breaking news: you prefer processed images, too. You cannot see unprocessed ones, anyway (not even by investigating the content of a RAW file with a hex editor).