StarAlignement, before or after DBE ?

[Julien]

Hi,

I want to apply a DBE on each of my subframe.
I opened one of my subframe, added samples, ... Now, I want to apply the correction on all subframes. For that I can use an ImageContainer and apply the DBE. But what if the subframes have a different orientation ? In the case of a rotation because of a meridian flip I think a FastRotation of 180° will be fine.
But how to handle the case of position difference due to dithering, pointing difference between two imaging session, etc ? If there are a lot of stars, it is difficult to put samples that contains no stars on each subframe.
So, I think to do the starAlignment before the DBE. But is it a good idea ?

Regards

Julien

[ngc1535]

Julien,

Why apply DBE to your subframes? The sky and faint gradients or flat errors are a form of signal. You need an integrated image to detect/characterize/model them. Attempting to do this on subframes for most objects (deep sky objects) will just introduce DBE modeling error noise (for lack of a better term).

Have you tried to create a combined image and apply DBE to it?

sincerely,
adam

[pfile]

just FYI DBE does not work with ImageContainer - only ABE does.

while what adam says is true if you have severe light pollution gradients what you are proposing could work, as the gradient will be clearly visible in each sub. however since you are going to apply ABE with the same parameters to each image, depending on the gradients, those parameters may not be applicable to each subframe, and you may end up under- or over- subtracting the gradient. i'd say it's more of an advanced, desperate technique and it's better to first try what adam proposes.

rob

[Niall Saunders]

Hi Julien,

What is your  reasoning behind the decision to increase your work load by some (large) linear factor whilst at the same time degrading final image quality by some non-linear factor?

I certainly cannot see what you hope to achieve.

[wadeh237]

If you have complex gradients that are significantly different between sub exposures (say, due to passing thin clouds), the LocalNormalization might be a better approach.

On a recent project, I had 3 nights worth of data.  On two of the nights, the OAG was rotated just a bit so that the shadow of the pick off prism was just visible in a corner.  And on one of those two nights, there were thin clouds passing through.  The data was good enough that I didn't want to throw them away, so I wanted to do something with them to mitigate the uneven illumination.  On the third night, it was crystal clear all night and the OAG was properly placed.

So what I did was pick the ten best exposures for each channel from the good night (I shoot short exposures with a CMOS camera, so 30 subs is not a whole lot of exposure).  I then integrated them so that I had a single master for each channel made up of the 10 good frames.  I then did DBE on each of those masters and used them for the LocalNormlization references when I calibrated the entire set of data.

I worked amazingly well.  I was thrilled with the results.