DBE tune up please...

[dhalliday]

Since I finally have abandoned ABE...I have some issues with DBE...
Here is the main (current) one,working with M1 data...
I get these black areas adjacent to M1..not good...
http://www.flickr.com/photos/daveh56/4336938680/
Solution...?

Also,if I MANUALLY add extra points...does DBE use them,since it did not add them in the first place..?
Also when it is laying points all over my M1...I can remove them,but what a pain..
what setting to change?
What is the impact of the size of the sample boxes..?
Here is M1 so far...wish it was not purple..!!
http://www.flickr.com/photos/daveh56/4335943023/sizes/l/in/photostream/
Its cold as hell here...literally..! 
Amazing how bad seeing was last nite,despite a clear sky...Could not get M51..even BINNED,with SIX minute exposures...which is wierd.

Dave

[Harry page]

Hi

How things change in a few weeks , now you are fussing over tiny little bits 

If highly stretch a image the background will usually have some problems  but in answer to your questions


1)  you can avoid holes by not putting samples near bright stars or on top of and or extra one near the areas giving problems

2) Dbe will use all the samples put on the image and extra ones are sometime desirable and some people do not use the automatic sample generation and put them in by hand

3) you can try and adjust the sample weight up to reduce the number of samples  ( even though you will be lucky if it does it in one go )

4) I try and select a reasonable box size so each sample can see some background ( but not to much ) and I tend to use between 10 to 14 depending on your image size

5) If you do not have a lot of stars in your image to use CC , I am afraid you will have to play with the colour manually 


And as for clear skies what are they

Harrt

[Silvercup]

Hi Dave:

I agree with Harry's comments.
Althoght image background has some problems you can do properly DBE if you don't put samples over stars or structures, just over background. I prefer do it manually. It's better a few manual sample than a hundred automatic samples.

If you have dark areas in the original image you can do DBE putting samples over these dark areas an a few samples on each image corner. Then substract DBE  flat image several times till image is corrected.

This is an example:

http://www.flickr.com/photos/47314701@N07/sets/72157623245998095/

Best regards. Silvercup 

[dhalliday]

Scup;
(do you mind my shorthand...?  )
Thanks for that,and welcome to Flickr...its a cheap way to leave images somewhere...
I looked at that +++ but my sleep deprived brain...?
And pixel math...

So you do SEVERAL iterations of DBE...?
Why not just have DBE do the subtraction ??
Could you add a little more text to the images..(for dummies..)
You have a great final result..?
Can you comment a bit on the noise level in this stack..?
Oddly when I play around with stacking with 1/2/20 darks I see no real improvement..
I have not yet done my homework on Pix's stacking algorithm...
I will add manually +++ and play around with location...and multiple iterations

Harry; Isn't it always the way..??...sorry for your sky..work overtime for a while...
Thanks,both of you..!

Dave

[dhalliday]

PS
I have also JUST (!!!  )
discovered that GreyCstoration works (roughly) 100 times better on LINEAR data...
!!!!!!!!!!!!!!!!!!!

Dave

[Jack Harvey]

Dave that may be true but I always save my noise reduction as a last or nearly last step in my processing so not as to lose data.  I find it works well there too with data that is no longer linear.  BTW I just discovered that if you use Atrous Wavelets and leave everything at defaults and check the K Sigma box and try it on a preview that includes a noisy background area you can de noise pretty well with this tool too.

[dhalliday]

Jack
Thanks...I will give it a go...
Noise...

Dave