DYNAMIC VERSUS AUTO BACKGROUND EXTRACTION

[ballyhoo]

Hello,

I Was hoping someone might know, to what extent can the auto iteration of the exaction process can accomplish what the more complex dynamic version can?

Also, the Primer suggests that subtraction is generally the way to go, but for vignetting , division is good. In my case, I probably have some sky-glow to deal with but I do not use a field flattner yet for my refractor.

Any suggestions? thank you!

[chris.bailey]

DBE EVERY time! ABE makes the decisions for you and more often than not your instincts are better.

If the gradients are a flat fielding issue (and many are) then division otherwise subtraction.

[jlodge]

I'm a PI newbie, I have found that in some cases ABE works a lot better than DBE, my understanding from Warren Keller's book is that ABE is preferable when there is a lot of background sky available, whereas DBE is better where there is a lot of nebulosity, so far my experiences confirm this.

[Bobinius]

I'm a PI newbie, I have found that in some cases ABE works a lot better than DBE, my understanding from Warren Keller's book is that ABE is preferable when there is a lot of background sky available, whereas DBE is better where there is a lot of nebulosity, so far my experiences confirm this.

I confirm that. ABE works really well when your target is a galaxy and you have plenty of uniform sky available. I just shot M101 without an antipollution filter, applied DBE using Generate in order to have lots of samples available and it did not remove the glow completely. Applied ABE after and everything disappeared (I mean the gradient  ).

Most of the time when ABE is applied to complex structures like nebulae with variation in dust brightness which can be close to sky glow/background the results are most of the time catastrophic (lots of parameter testing needed to differentiate between the two), imho. DBE allows you to really see and control what's happening.

[jlodge]

An update on my experiences, I am processing some RGB of the Leo Triplet and I have bad gradients, so I've done one run of DBE first with a minimal amount of fairly large sample points in the corners, on the sides and a scant few in the middle, this worked pretty well. Then I did another run with ABE and this seems to have evened things out a lot.