Mosaic SNR Differences?

[james423896]

Attempting a mosaic to prove out the process and I cannot get a seamless join, it isnt even close with the SNR looking very different. I collected 6x5m Ha subs for each pane, and a total of 2 panes. I then calibrated and integrated the subs to produce a couple of files.
part1_ha.xisf
part2_ha.xisf

Following the Steve Allan youtube tutorial I first created a mosaic template, and I then registered the individual panels to that mosaic producing
part1_ha_r.xisf
part2_ha_r.xisf

I then used GradientMergeMosaic on these files to produce the MergeMosaic file. HaMergeMosaic.xisf

This shows a clear difference between the panels, nothing like a seemless mosaic! What am I doing wrong?

Link to files here:
https://1drv.ms/u/s!AtooZYq7lKxzgUMS-agAV1zNXOrn?e=6aqEbO



TIA,

[pfile]

does that tutorial include the use of DNALinearFit? if not, that's probably what you need.

this is an add-on script; if you google it you'll find it.

i havent looked at your frames but 6x5m could be so short that the SNR is not really that great in either pane.

rob

[Juan Conejero]

This mosaic builds perfectly in PixInsight 1.8.7 with the latest version of the StarAlignment tool; see the attached thumbnail.

All you need to do is using thin plate spline registration models with distortion correction enabled, frame adaptation enabled, and the register/union mosaic mode. No other tool is required. You have a potentially fantastic mosaic here. I encourage you to repeat the whole image preprocessing task with the original data, especially image integration to remove all hot pixels and plane trails that are present in the individual mosaic frames.

Although this six-frame mosaic can be built without problems with StarAlignment, the new MosaicGenerator tool (to be released before the end of the year) will be faster, more accurate and capable to produce seamless large mosaics like this one. The new tool performs astrometry-based image registration, in a similar way to the MosaicByCoordinates script. You can also use the MBC script to generate this mosaic right now.

[dave_galera]

Juan, So are you saying that GradientMergeMosaic is now obsolete and no longer necessary?

[Juan Conejero]

Hi Dave,

No, not at all. It's just that the new version of the StarAlignment tool is much better at merging mosaic frames, thanks to a new randomized image combination algorithm that minimizes visible texture differences.

[james423896]

Don't think those replies were for my mosaic?

I got to the bottom of this one although I don't understand the root-cause. Frame-adaptation was causing the SNR differences, running without this lead to smooth stitching...