View Single Post
Old 04-06-2016, 01:50 PM
Shiraz's Avatar
Shiraz (Ray)
Registered User

Shiraz is offline
Join Date: Apr 2010
Location: ardrossan south australia
Posts: 4,821
Originally Posted by gregbradley View Post
Hi Ray,

I am not sure how you concluded that from Craig's article. The gain and offset had a procedure to be set to keep the resulting images within the 65535 levels that 16 bit gives. Its not dependent on exposure length but setting what the sensor will record without loss of data.

So you are saying shorter exposures at higher gain may clip dynamic range but you get better read noise with the higher gain settings?

Just out of interest what are the read noise values for this camera at different gain settings. is that known? I have read figures from 1.6 to 3.6.
Which by the way in CMOS world is quite high. I should look for it but I did read a test of the Sony A7r2 BSI 42.4mp full frame sensor read noise at some ISO levels as .3 electrons. This makes sense as OMD cameras are often reported by owners as noisy even at low ISO. The point here is not to take anything away from the ASI1600 as its a new paradigm and way cleaner than CCDs but more imagine one of those Sony full frame BSI babies in a cooled body like the ASI? Wow.

But this camera is 12 bit? Or is it 12 bit output but 16bit internal processing? (For example Sony A7 cameras output is 12 bit sometimes depending on features activated but the processing pipeline is 14 bit).

How many levels is 12 bit is 4096 levels of brightness.14 bit is 16384 levels of brightness.

Yep, takes a bit to get the head around, but the read noise decreases as the gain increases. With a CCD the signal to read noise ratio stays much the same as the gain changes - CMOS are completely different beasts in that there is no single "optimum" setting for the 1600, whereas there is for a CCD.

It is all fully documented - the read noise can be as low as ~1.2 electrons, but you have only 9 bits dynamic range at that gain setting. Sony chips will do the same thing - pity there are no large mono Sony CMOS chips. I think I read somewhere that Sony make about 800,000 image sensors a day, so they may not be all that interested in changing a production line to do a run of a few hundred mono DSLR sensors - I think that it is amazing that ZWO managed to get Panasonic to do it.

The camera has a 12 bit ADC. You end up with 16 bit numbers in the images, so I assume that the software pads out the bottom 4 bits. This is the same as for the 694, except that the 694 has random noise in the bottom 4 bits rather than zeros. In terms of image quality, no practical difference at all - since random noise and zeros are both useless.

Last edited by Shiraz; 04-06-2016 at 04:20 PM.
Reply With Quote