Basic PixInsight Preproc Lights Calibration - mottle from Master Flats or Darks?

[AnakChan]

Hi,

Just seeking some assistance with some basic PI preprocessing. This is from a OSC/DSLR. ISO2000/300sec subs.

I noticed that in generating my lights calibration, that if I have both may Master Darks and Master Flats, I get a somewhat mottled background. However if I select only one or the other, then it's much smoother.





In generating my Master Darks, I've pretty much only used straight ImageIntegration NoNormalization/WinsdorisedClipping from the RAW files (not calibs). And my Master Flat from Flat Calib off master Dark Flat. I've also tried the more traditional LightVortexAstronomy with Master & SuperBias, DarkCalibs/MasterDark, FlatCalibs/MasterFlat but they come out the same with noisy background. I've also tried different rejection methods, SigmaClipping, WindsorisedSigmaClipping, & LinearFitClipping for the various master files but to no avail.

Is it to be expected that when one uses a both master dark/flats that the background starts to look more mottled?

[pfile]

i think you should debayer the lights before evaluating this... the pattern you see might just be caused by the flattening emphasizing one channel more than the others.

rob

[AnakChan]

Thank you Rob. Not certain if I've done this correctly?

[pfile]

probably correct - a green cast is not unusual in OSC cameras when the STF channels are locked. although the mk.1 eyeball is not so good at evaluating the noise in images, i don't see anything really wrong with it.

anyway one thing to consider is that ISO2000 may not be a good choice as it is probably not a pure 'analog' gain setting. pure analog ISOs are similar to powers of 2 - 100, 200, 400, 800, 1600, 3200, etc. there's not much point in running at the 'inbetween' ISOs since the sensor is configured the same as the next lower ISO and then a digital gain (digital multiplication) is applied to the data coming off of the ADC. even if the data is brighter, it's an illusion and nothing that can't be done in post-processing anyway.

rob

[AnakChan]

Hi Rob,

I’m using ISO2000 cos on my A7S, the read noise drops somewhere between ISO1600-2000. That seems to be the recommended ISO by other A7S Astro owners.

No concerns about the cast. I can clear that up in post. Does the mottle “acceptable”? I should have debeyered the others too for a comparison.

On CN forums, someone suggested the noise evaluation script across all 3 for a more calculated comparison rather than just visual comparison. I’ll see if I can try that later today.

[pfile]

OK i thought it was a canon DSLR for some reason. sony is a different beast...

yes you can use NoiseEvaluation but juan has cautioned against using NE to compare across different images - if the two images are not scaled to one another the results don't mean anything. here is one post where he describes how to compare the noise between two images:

https://pixinsight.com/forum/index.php?topic=8434.0

at the end of this thread juan published a modified NE script that will scale the input image first so that you can compare different images with the script.

https://pixinsight.com/forum/index.php?topic=13715.msg82701#msg82701

i assume what you'd want to do is compare an uncalibrated image and a calibrated image using this version of the script to see if the calibration has somehow degraded your subexposure.

rob

[AnakChan]

Cheers for the pointers Rob.

All the 3 samples are from the same raw image. In fact the way I produced the 3 diff samples was merely in the ImageCalibration and selecting/unselecting the Master Flats/Master Darks and global applying.

[pfile]

right, but i'm not sure the files will be scaled identically after calibration with flats. at the very least it would be a good test to see what the noise evaluation script says against the unchanged files and what it says after they have been rescaled to one another.

rob

[AnakChan]

I initially posted the Noise Estimator prior to debayering and was told that it was nonsense. I'd have to debayer it first, then PCC correct the colour balance and here's what I've got now.

1) Light sub calibrated out Master Dark and Master Flat

* Channel #0
?R = 2.708e-04, N = 5951011 (49.10%), J = 4
* Channel #1
?G = 2.009e-04, N = 7935125 (65.47%), J = 4
* Channel #2
?B = 2.178e-04, N = 4394168 (36.25%), J = 4

2) Light sub calibrated out Master Dark only

* Channel #0
* Channel #0
?R = 7.142e-04, N = 7184810 (59.28%), J = 4
* Channel #1
?G = 4.419e-04, N = 7366214 (60.77%), J = 4
* Channel #2
?B = 6.076e-04, N = 6227397 (51.38%), J = 4