Adding Data to Existing

[Farzad_k]

Hello,

I am just curious as to how the workflow will be when data is being added to the data that has already been processed. I intend to go out and collect more data on an object that I already processed.

I know that if I keep the same exposure length and the same ISO, and if the weather is about the same, then I won't have to worry about dark frames and biases.

In PI, the Pre-Processing workflow as I understand it is:

[1] integrate darks
[2] integrate biases
[3] integrate flats (if any)
[4] calibrate lights using integrated darks, biases and flats
[5] register (align) calibrated lights
[6] integrate (stack) registered calibrated lights

Where would the new data (light frames) be incorporated? Or do I have to do step [4] again.


Thanks

[pfile]

if your optical train has not changed, you can start from #4 with the same masters on the new lights, then register the new lights, and then run a new integration that includes both the old and new lights. the old calibrated and registered lights are still good to use, as long as you use the same registration reference for the new lights.

if the optical train has changed, then you should get new flat subs, calibrate/integrate them, and then use the new master flat instead of the old one in step #4.

rob

[Farzad_k]

Thanks for the input.

Farzad

[Niall Saunders]

Hi Farzad,

[1] integrate darks
[2] integrate biases
[3] integrate flats (if any)
[4] calibrate lights using integrated darks, biases and flats
[5] register (align) calibrated lights
[6] integrate (stack) registered calibrated lights

At #1, you might only need new Darks if your image acquisition parameters have changed (Exposure time, CCD Temperature, Gain, Offset, etc.).

At #2, you really only need new biases if you are using significantly different parameters, or equipment, to gather your data - if you are sticking with, for example, a CCD imager, then it is unlikely that Biases will change (unless you have changed Gain and Offset parameters) - temperature and exposure time are pretty meaningless for Biases. But, if you haven't taken Biases in several months, it might be worthwhile creating a new MasterBias - just to see if your CCD behavious is changing 'with age'.

Remember - at #3, you really want to be 'calibrating' your flats, as well as 'integrating' them. This means that you may well need to consider acquiring Darks to suit your Flats as well (you would not need a new set of Biases). You can check the need for this by comparing Darks taken at your exposure time for your Flats, versus Darks taken at your exposure time for your Lights. Calibrate and Integrate a set of each of these and then look to see how the Masters compare. If your exposure time for FlatDarks is short enough, then a MasterFlatDark may not be that different from a MasterBias - in which case you could simply use the latter instead - this will be camera-dependent.

Rob is right about #4 - you need to think about 'how' the new data has been acquired. This will determine whether you can use old calibration data, or whether you realise that new data would be more appropriate.

At #5, when you align your newly-calibrated Lights, you will need to think about what you are going to use as the 'reference' image - the same reference frame from the original set, or the fully-calibrated and stacked image from the original set? Perhaps you still have all of the original, calibrated, Lights - in that case you may want to throw ALL of the data into the StarAlignment Process, and see what comes out.

At #6, ypu have a similar decision process as you had at #5. You could integrate the new data totally separate from the old data, and then 'merge' the two MasterLights using the likes of PixelMath. Or, you could include the first MasterLight into the set of new data - and ImageIntegrate all of the images together (this may not give you the best result). Perhaps the 'best' method - if you still had the original calibrated Lights, would be to ImageIntegrate the complete data set. This would give you, and PixInsight, the most statistically viable data set with which to perform the integration and noise rejection - but, as always, you should consider trying all the different approaches to see what works best for your data.

In terms of the pure mathematics and statistical analysis involved in the whole process, it could be argues that there should not be any 'shortcuts' - but not everyone has the time, or inclination, to hammer their data through the full calibration procedure every time.