Am I running into Mosaic Limits?

[Seldom]

Are there limitations to using "Register Union Mosaic" or "Register Union Separate" tto create a mosaic?

I'd like to create a 3 tile x 3 tile mosaic of SH2-240 (Spaghetti Nebula) but I'm get "computing intersection" failures when I try to create the mosaic using either "Register Union Mosaic" or "Register Union Separate". Failures usually occur when I have 5 or 6 tiles combined in "Register Union Mosaic" or when I try to combine one set of 3 tiles (combined) and one set of 6 tiles (combined)in "Register Union Separate". The star detection reports about 76000 (slightly more than half the total) stars before it fails.

I've successfully combined all the tiles twice, so I assume they are good, but I've had numerous failures using the same data I was able to combine. 

My tiles are 4656 * 3520 (from an ASI1600MMC).  My computer is a 64 bit Windows 10 desktop with an i7-6700 CPU and 16 GB of RAM.  The mosaic was created using TheSkyX's Mosaic Tool.  The images were saved by TSX as 16 bit .FIT files.  I calibrated and stacked them in PI as 32 bit .XISF files. Combined mosaics of tiles are saved as 64 bit XISFs.

Thanks in advance.

[pfile]

yes as the mosaic gets bigger i find that i have to put previews around the stars in the mosaic which abut the frame i'm trying to bring in.

however, i would also say that i haven't made a mosaic like you are doing in years and years. if you use ImageSolver on each pane and then save the files to disk, you can use MosaicByCoordinates to generate the registered panes. assuming you can get ImageSolver to work (which is pretty foolproof these days) then it is a heck of a lot easier than how you are doing it.

rob

[Seldom]

Thanks, pfile.  MosaicByCoordinates is a lot faster than.  "Register Union Separate".   But I've been using "dnalinearfit" to even out my tiles, and dna  linearfit isn't playing nicely with MosaicByCoordinates.  If I apply a dnalinearfit after a MosaicByCoordinates, I find a bunch of tiles go blank.

[pfile]

this can be caused by ragged edges in the mosaic which have low SNR. check that you have cropped the input panes well.

rob

[Juan Conejero]

The new MosaicGenerator tool, which I hope to release during the coming months, will be a PCL-based, largely improved version of the MosaicByCoordinates script. It will make mosaic construction a fast, efficient and accurate process in PixInsight, leveraging the outstanding accuracy of our astrometric solutions for local distortion correction.

MosaicGenerator, along with DynamicBackground, the successor to DBE, should have been released around the end of 2019, but a number of unexpected issues and emerging projects have prevented me from investing the required time. I hope this will change soon.