Hi Harry, Simon,
I assume that you have at least done some form of analysis on your camera to 'prove' that there is no residual 'noise' element in your thermal 'DarkCurrent'?
To do this I would capture, say, around 10 normal long-exposure Darks - at whatever temperature you normally run the camera at. Then I would take a similar number of Biases, at the same temperature.
I would then want to graph the following information (one set acquired for each image) as y-values, against an x-axis running from 1 to 20, representing all 20 images :-
Max ADU
Min ADU
Mean ADU
Median ADU
SD ADU
Perhaps a picture would explain things better - the following is just an example from a test I did recently where I compared FIVE different 'image groups'
In this image, the x-axis represents a total of over 200 images, whose statistical data has been analysed using the excellent "Series Analysis" tool in AIP4WIN (Juan, please note
)
These 213 images represent five individual data types:-
The first group contains 30-odd Darks taken at 7.5C and with 3-sec exposures
The second group contains 30-odd Darks taken at 7.5C but with 2-sec exposures
The third group contains 30-odd Darks taken at 7.5C and with 4-sec exposures
The fourth group contains 60-odd Biases taken at 7.5C and with 0.0001-sec exposures
The fifth group also contains 60-odd Biases taken at 7.5C and with 0.0001-sec exposures - but taken two days later
(see how the Bias 'mean' level has shifted?)(BTW - ignore the '7.5C' temperature - the thermal probe is no longer making good contact with the CCD. I am confident that all of these images were perfectly temperature regulated at -1.0C)
As far as the four graphs are concerned, the first shows three superimposed graphs for Mean, Median and Mmh(??) ADU (or PV values as they are called in AIP4WIN)
The next one down shows the Min ADU values for all the images
The third one shows the Max ADU values
The fourth one shows the StdDev ADU values
The point I am trying to make is that, for you to be able ('statistically') to use Biases 'instead of' Darks, then you should IN NO WAY be able to detect where YOUR group of Biases ends and YOUR group of Darks starts - unlike the 'step changes' that are visible in my example (even in my quick'n'nasty example, it is quite easy to detect the small step-change in Mean/Median values when we move from the 2-sec Darks to the 4-sec Darks - corresponding to the FAR more noticeable step-change in the Max and StdDev ADU graphs below).
Of course, your
actual experience counts for a heck of a lot more than 'statistics'. If it works, it works
But, Mr. PixelPoliceman
would always want you to have at least 'thought' about what might ACTUALLY be going on 'behind the scenes'
(And I hope I haven't just spent 30 minutes typing drivel - I 'think' I know what I was trying to point out, but I can't really remember any more. "Time for bed", said Zebedee)
Cheers,
