Calibration (using already constructed 32-bit IEEE Float files)

[ngc1535]

Hi,

Thank you for letting me participate in this forum. I understand that many of my initial questions are probably covered in various FAQs... but it is the promise of being able to ask follow-up questions to my initial special inquiries that I think will serve me well. So here goes.

Yesterday I attempted to calibrate a set of data using Darks, Flats, and Biases that were created outside of PI using another software. (I do not have PI installed on the machine that acquires the data since I do not process on this machine with the exception of making master calibration files. This saves me from having to download the individual files that I really do not need.)

The calibration files are 32-bit IEEE float format and undoubtedly need to be rescaled to make PI happy and get the expected results.

I have attached an image of the ImageCalibration utility to see my settings.
I checked the "calibrate" dark frame since the bias needs to be removed to create the thermal frame.
I did not check "calibrate" for the flats because the bias has already been subtracted.
Optimize is selected since my darks are 30 minutes and my data is 15 minutes in length.

I look forward to some help!
Thanks-
Adam

[Juan Conejero]

Hi Adam,

... calibrate a set of data using Darks, Flats, and Biases that were created outside of PI using another software.

In general, that won't work. The main problem is that the FITS format does not define any standard way to describe the numeric range to which floating point data are referred. This has been discussed many times on this forum. Each application follows its own rules, which are in most cases undocumented. In the case of PixInsight, floating point images are always stored in the [0,1] range, where 0=black and 1=white. However, with the sole exception (AFAIK) of DeepSkyStacker, the rest of applications don't follow this rule. Without a common numeric range of reference, calibration masters lose their physical meaning and hence are unusable.

You can try with input hints to adapt the alien data to PixInsight. Depending on the other application, this may or may not work. Input hints are special control words that modify the way file format modules work in PixInsight. For example, try the following:

upper-range 65535

which should work if the other application stores floating point data using a 16-bit integer range.

[ngc1535]

Hi Juan,

I have attempted to follow your advice by putting "upper-range 65535" in the input hint field.
Unfortunately it did not work. There are more important things for me to understand than the vagaries of non-standard format issues.
Thus I must reveal my specific case (no doubt you surmised it)- I used CCDStack to create the master frames.

1. The darks are created from a set of 30+ images that are min/max clipped by 3-4 values and then the mean is returned.
2. The same as the above for the biases (usually from 50+ measurements).
3. The flats are are simple averages (means) of images (with the master-bias subtracted).

The original data is 16-bit integer. The range therefore, as you correctly state, will have the upper limit of 65,535. However, it will include fractional
values.

Any help in getting my data to calibrate would be great. Did you identify anything in the fields that I populated that might not be logically correct? (Were you able to see the attached image to the original post?).

Thanks,
Adam

[Andres.Pozo]

Any help in getting my data to calibrate would be great.

Don't use the masters generated by CCDStack. You have to generate them from the original calibration shots using PixInsight.