Mosaic Issues in Astro Pixel Processor 2.0.0 beta29

I have attempted to reduce the number of frames stacked to generate a mosaic panel - 8 subs with generous overlap.  I am still getting the same or very similar error.  I could really use some assistance in navigating this issue.  Thanks in advance.

How many CPUs is APP showing on the slider at the top left?

@walsc It was either 31 or 32 CPUs.

This should be fine. What are the parameters on you ticked or changed for this mosaic?

@walsc There are quite a few parameters to tic.  Here is where I currently stand.  I have managed to create 4 FITS files in APP that together should create my Milky Way panorama.  They were created from RAW files, Master Darks, Master Bias, and Bad Pixel Map.  I then restarted APP, loaded in these 4 FITS files and proceeded to stack them in Mosaic mode.  They loaded fine. I used Calibration defaults. Analyze Stars went fine, no signs of anything radically different from the results when building the mosaic panels.  When I ran registration, I received the error message below, which I can only speculate as to the cause.  Thanks for your engagement.

Regards, Martin

P.S. It was 32 CPUs

Registration Tab Settings:

Pattern Recognition:  quadrilaterals

Scale Start: 1      

Scale Stop: 15

Flip Descriptors in X/Y: unchecked

Dynamic Distortion Correction:  Use DDC checked; Same camera and optics not checked due to message from APP indicating not to do so when using previously calibrated files.

No Distortion Model

Distortion Margin: 0.005 (default),   Save Distortion Model unchecked

Registration Mode: mosaic

Registration Model: Calibrated Projective

Rotation: 000 degrees

Calibrated Projective Options

Projection Method: equirectangularProjection

Move COP Horizontal: 000

Move COP Vertical: 000

Save Registered Frames:

Scale: 1.0

Interpolation Method: lanczos-3; no under/overshoot

Error Received Using Quadrilaterals

Error Received Using Triangles

Error Received Using Pentagons - a total non-starter

OK. Try this:
Tab 0,1,2 unchanged.
Tab3: 1000 stars, maybe more
Tab 4: Scale start: 1, scale stop:10, distortion correction on, same camera off, registration mode mosaic
Tab 5: normalization mode: advanced
Tab 6: enable MBB, slider somewhere between 10 and 25% (depending on overlap)
everything else default.

I did not include Tab 5 and 6 settings as the images will not register.  I will try >1000 stars in Tab 3.  Perhaps that will assist with getting past the Registration phase.

 I have been using max MBB in general.  It does not appear to slow my processing down much.  I have also been using LNC1, 3 passes.

I do mosaics quite a bit and MBB is set to 15%, LNC2 3 iterations, and for star analysis I find that 2000 seems to work best. Also make sure that on tab 5 normalization is set to advanced.

@imnewhere I usually do not use LNC, most of the time it is not necessary or does weird things, and it slows down the process significantly.

@martinontheroad As APP has strange limits everywhere, I do not recommend to set MBB at a very high level.

@walsc I will take that into consideration.  But, would settings in Tabs 5 and 6 impact Tab 4, star registration?  My instinct says “No,” but others may know better.  Right now, I am stuck at Tab 4 and cannot Register my mosaic panels.

Ultimately, we will not be spared the fact that sooner or later we will have to do everything with PI anyway, there is no longer any support from the developer here.

@walsc I have very good results with it when mixing data with high variability and CMOS with CCD data. I'm also running a monster of a computer so I am not worried about slowing down either. The current mosaic I am working on has 56 panels for each channel of SHO, so 168 total master lights in it.

@imnewhere Thanks for this.  I’ll give these settings a try. Maybe the 2000 stars will get me past the Registration tab.

@martinontheroad 

It definitely went through the whole process for me. Too bad APP will not combine them in RGB combine without me resizing them down

@imnewhere But you can register and combine these results in PI.

@walsc will PI combine the masters without redoing registration? I know that I have combined masters from PI in APP before without registering them. I may try running them through PI this evening. This is a monster size image though.

No. You must apply an astrometric solution (scripts/imageanalysis/imagesolver) for each master in PI, then you must register them with processes/imageregistration/staralignment. After that you can combine nthem with PI.

But if your masters are already registered with APP, maybe PI will combine these without further registration, you have to try.

They have already went through everything up through the integration tab, but APP isn't able to do the combine RGB due to the files being the size they are.

@imnewhere I would just try if PI can combine these.