Basic Calibration Question

[Adrien]

I have a new QSI 683 camera. When I view the Median Bias ADU Count of this camera in Both Maxim and CCDStack it shows the same ADU count of 259. When I view the same Bias frame in PI it shows count of K:0.0037 (not sure what that means).
When I calibrate the individual Bias frames using Vincent's tutorial word for word with the Image Integration Module. I end up with a Master Bias that has a ADU count of Zero, with is sort of unusable. I have the same problem with the Master Dark as well. My work-around is to calibrate using CCDStack and then continue processing with PI.
I do ensure that I choose No normalization in the Image integration module, as per the tutorial
Appreciate any assistance to see were I am going wrong.
Thank you
Adrien Richardson

[Nocturnal]

I don't know why your normalized bias would be all zeros but the 0.0037 is simply a 0-1 float representation of your integer frames. to get the median you'll first have to run the statistics process. I would also prefer if PI would leave files in their native format but that's a battle long ago lost.

[Adrien]

Thanks for your reply. To assist solving the problem my setting are detailed below:
ImageIntegration:
   No hints
   Image Integration
      Average
      No normalization
      Don't care (all weights=1)
      Scale estimator (default Iterative k-sigma)
      Select generate integrated image
      NO (Evaluate noise)

   Pixel Rejection (1)
      Winsorized Sigma
      No normalisation
      all boxes selected

   Pixel Rejection (2)
      Min/Max low: 1
      Min/Max high: 1
      Percentile low: 0.020
      Percentile high: 0.010
      Sigma low: 4
      Sigma high: 3
      Linear fit low: 5
      Linear fit high: 2.5
      Range low: 0
      Range high: 0.98

Can anyone see why the resulting normalized Master Bias would be at Zero ADU? The RAW files are captured by Maxim, there has been no pre-processing on them, simply captured then downloaded.
Thanks for any assistance.
Adrien Richardson

[Adrien]

Just to add, the FITS can be downloaded for review at the following link:

https://www.dropbox.com/sh/a83zrlsqfo6c084/iWP31nt5pE

Thanks
Adrien Richardson

[Adrien]

Can anyone please help?
Thank you

Adrien Richardson

[Geoff]

Hi Adrien
Please upload some of the unprocessed bias files.
Geoff

[mschuster]

Adrien,

The master bias you posted is not Zero ADU, I measure the median ADU for both the single bias and the master at about 260 ADU using the Statistics process. The master looks good to me, with a cursory look, much as I would expect for a KAF-8300. I have a QSI 683 also. Maybe you posted the wrong master?

Also, the K:0.0037 can be changed to read in ADU easily. At the top of the Statistics window, choose a 16-bit range.

Thanks,
Mike

[Adrien]

Hello Mike
Thanks so much for your assistance. You are indeed correct. I selected 16bit, and now I see about 260 ADU.
Being new to PI what confused me is that when I process with the pre-processing script, I keep getting the error “No correlation between the Master dark and target frames”. So I thought to go back to basics, and try to calibrate Bias / Dark / Flat separately to see where the problem is being introduced. This is where I had the K:0.0037 issue, PLUS both CCDStack and Maxim agree that the Mean ADU count is Zero. I guess the information contained in the PI FITS header is unique to PI and not compatible with other programs.
Is there a way to make the Readout options / integer / 16 bit selection a preference setting? It keeps reverting back to defaults showing the K:0.0037 every time I restart PI.
Once again Thank you!

Adrien

[mschuster]

Hi Adrien,

Regarding the “No correlation between the Master dark and target frames” warning, PI is trying to scale the dark master in an attempt to minimize noise in the output calibrated frame (i.e. to match temperature and/or exposure time, etc., of the dark master to the target frame). If the resulting scaling coefficient is near zero then the warning is displayed. I have seen this warning when I tried to use a 2400 second dark master with 24 second panel flat frames. The scaling coefficient was usually near 0.01 (small) but sometimes it was even smaller and I got the warning. I now use a 24 second dark master for my flats to avoid the problem. IMO the warning indicates a problem that should be fixed somehow.

Usually my version of PI on Win7 remembers the range preference setting. But sometimes when I launch PI the preference is forgotten. Also sometimes the tool bar settings are forgotten. And also sometimes the scripts that I have added to the script menu are lost. I have no idea why. Most the time everything is remembered, but sometimes not. This problem has been occurring for over a year with all of the versions of PI that I have tried. It may be that my antivirus software is inhibiting system file updates occasionally.

Thanks,
Mike

[Juan Conejero]

I would also prefer if PI would leave files in their native format but that's a battle long ago lost.

There's no battle to win here. PI leaves *all* files in their native format: it fully supports 8, 16 and 32-bit integer and 32, 64-bit floating point images. However, the representation of pixel sample values is virtual in PixInsight. This means that you can inspect pixel values in any range, irrespective of the actual data type. For example, you can have a 16-bit image and read its pixels in 8-bit format if you want. The default readout format is the [0,1] floating point range. Mike has already explained how you can use the ReadoutOptions tool to select the representation you prefer.

[Nocturnal]

No, that is not what I meant. If I load a file with integers PI should show integers in readout and not rescale to reals. I know I can change the readout options but it's a pain and not intuitive so I don't mess with it unless I -have to-.

A lost battle indeed.