Yet Another DBE-question

[bianry]

I finally got some RGB from my white whale NGC 5033.
In the green channel (pic 1) there is a very visible gradient.

Okay, DBE to the rescue. There seems to be two schools on how to place the samples. One says I should place a few and other says I should plaster the image with them but avoiding stars and galaxies/nebulas.

If I have a gradient like this one. With a brighter part in lower left quadrant. How do I place the the samples? On the brighter part mostly? Or some there and some on the darker parts?

With this one I get the best result if I plaster the thing with samples. Which is basically what AutomaticBackgroundExtraction seem to do.

I really, really miss the reference documentation on this function. And what is the deal with the symmetry checkboxes? When do I use them?

If someone else also don't have a life on a Sunday afternoon I would really appreciate some pointers. I have read a LOT of threads/tutorials on this function but most seem to say that I should try till it looks OK.
A bit of background info would help. What is happening under the hood?

regards

Mats

[bitli]

You may try to look at
http://www.harrysastroshed.com/pixuser/DBE.html
or otherwise explore with google...
For the number of points to use, you may want to try both approaches, see which one is nearer the solution, then work from that.  I tend to try first with a few points (say 8 x 6), large enough if there is enough background, just because it is less work. But I am not able to provide a general rule, unfortunately (I guess if there was one, it would be implemented in the tool already).
-- bitli

[bianry]

Thank you bitli. Yes, Harry is a great help. As is Google.  What I have not found any answer to is whether
you can `damage' the image by applying too much correction. It seems to get noisier but I read here that it is caused by reweiling the noise, so to speak.

R

Mats

[pfile]

i assume that the "less is more" strategy applies to situations where your LP gradient is "well behaved" and you have a more-or-less linear gradient across your image. in that case you really want to pick only a few samples. you probably only get that from dark skies.

for me though under very light polluted skies, as i track the object across the sky i end up with a "U" shaped gradient sometimes (if i am lucky) and if i am unlucky, a very complex gradient. in those cases i think you have to sample the image more densely for DBE to really identify where the edges of these gradients lie.

rob

[Geoff]

Another useful link
http://www.deepskycolors.com/archivo/2010/05/28/removing-gradients-while-preserving-ve.html
Geoff