How can I remove this gradient in my mosaic?

[dc_robert]

Hi all. Relatively new astrophotographer (< 1 year) and even newer to PI (< 3 months), so I'm still making lots of mistakes even as I try new things.  One of my recent experiments was to create a mosaic consisting of the North America and Pelican nebulae.

The problem I'm running into is a gradient between the two merged images that I don't know how to eliminate.  I've tried using DynamicBackgroundExtraction, but typically it only makes the gradient more noticeable (at least, with respect to STF Autostretch).  Here's what the linear image looks like following StarAlignment and GradientMergeMosaic + Autostretch (I'm largely following Kayron Mercieca's mosaic tutorial as best I can using my DSLR OSC images rather than his CCD mono ones):



Closeup:



Part of the problem may be that I have more acquisition time on Pelican, though it seems to me that can be factored out in PI.  Perhaps I need to tweak some parameters during StarAlignment or GradientMergeMosaic?

Here's a link to the linear fts file I'm trying to work from: https://dl.dropboxusercontent.com/u/24745/Pelican%2BNA_Nebula_02_mosaic_GMM2_Crop.fit

Thank you for your assistance!

[TobiasLindemann]

Hi,
do you have checked the box for "frame adaptation" in the star-alignment process?

Greetings
Tobias

[dc_robert]

Yes, did. Both for the initial StarAlignment merge (Register/Union Mosaic), and also when I ran it again as part of the process of ridding the merge line (Register/Match Images).

[JoLo]

Looks more like differences in processing between the two panels than a gradient to me.  For instance, the panel with the NA seems to have more noise than the Pelican panel, also brightness/contrast appears to vary.  Does this line follow where the two panels were merged?  Linear Fit can help equalize the background differences.  I often see gradients that are more gradual in transition, this seems to abrupt to me.

[dc_robert]

The only processing done was stacking each panel separately in DeepSkyStacker, crop, StarAlignment, and GradientMergeMosaic.  Could somewhere in these steps be the cause?

The difference in noise I think is accounted for by more data for the Pelican panel.

Yes, the line is where the two panels merged.  I'll look into Linear Fit.

[pfile]

i'm thinking that since these are OSC images you should probably color balance each pane first, or else split each pane into the 3 channels and create mosaic panels for each channel individually, then recombine them to RGB and color balance the mosaic'd result.

actually you should probably do DBE on each frame individually before doing anything else…

rob

[JoLo]

Whenever i do a mosaic I try to get "close" on brightness/constrast/color balance/noise on the individual stacked panels, linear fit is one tool that can help with this.  Once the panels are merged, you can fine tune the parameters using masks to get a better match and remove any obvious line where the panels merge.

I agree with Rob, I tend to do DBE on individual channels, as a gradient can vary between color channels.  I do this on individual filters from a CCD camera, but as he suggests an OSC can be split, dealt with individually, then recombined.

Mosaics have their unique challenges and get better with experience.

BTW, very nice image, Mr. Pelican can be tough.

Joe

[dc_robert]

Thank you all for the advice and kind comments. I'll give this a try!