15th Feb 2024: Astro Pixel Processor 2.0.0-beta29 released - macOS native File Chooser, macOS CMD-Q fixed, read-only Fits on network fixed and other bug fixes
7th December 2023: added payment option Alipay to purchase Astro Pixel Processor from China, Hong Kong, Macau, Taiwan, Korea, Japan and other countries where Alipay is used.
Hi Mabula
I'm working on a NGC6888 12 panel mosaic and I finished gathering one frame per each panel.
I followed you and Sara's tutorial and APP in general did a great job but there is one border where the alignment is not perfect.
I'm wondering what parameter I should adjust to make it perfect.
Anyway here is the results with a quick process.
This is a just quick test image before I start capturing more frames.
12 panel / resolution 15086x8870
1 x HA, 1 x Oiii per panel
30 minutes per frame
http://www.astrobin.com/full/302613/0/
Hi Dong Hun,
Since I don't know the exact settings that you have used now, I'll give you some general tips on how to possibly improve this:
- Try to get at much stars detected in the images in 3) analyse stars, using "detect above noise level" kappa and increase the "limit #stars" value. The more stars that are detected, the more star pairs can be used in calculating the exact mosaic.
- in 4) register, make sure you have scale stop increased to 10 instead of 5 (default). 10 would correspong to 10 % overlap between the panels. (A scale stop of 8 would be 100/8 = 12,5 %)
- in 4) Register enable distortion correction
- in 4) Register use projective registration model with mosaic mode
- in 4) try with "same camera and optics" on and off, in case of off, increase number of overlaps possibly to 6 from 3.
Let me know if this helps 😉
Your image on astrobin is already awesome 😉 I think...
Cheers,
Mabula
Thank you Mabula.
I've got a perfect 12 panel mosaic now!
I did multiple things so I'm not sure what exactly fixed it.
Here are my successful settings
I have 2.33 arcsec/pixel.
Analyze stars:
Min star size - 4
clip star profile 0.05
filter star profile enabled
kappa - 4
limit # of stars - 1500
Register
triangle (this was must)
scale start - 5
scale stop - 15
Enable use dynamic distortion correction
Enable same camera and optics
registration mode - mosaic
registration model - projective
Donghun
Excellent Donghun !
That's great to hear... 😉
Your working settings seem to indicate that overlap between some of the panels was rather small. Is that indeed the case? (Can't find the seams so I don't know...)
The triangles are needed in case of very little star pair matches between some of the frames. So triangles, in that case, are more robust than quadrilaterals.
It might have been resolved with a higher @limit of stars (2500-3000), but if you are already detecting all stars in the regions of overlap than this will add nothing.
The scale stop of 15, means that the registration engine tries to match patterns (of triangles in this case) as small as 1/15th of your individual panel dimensions.
If your panels were already created with distortion correction in the registration engine, disabling "same camera and optics" could also give you a better result.
But it depends, For larger mosaics you do need to tweak the settings here and there.
Really looking forward to the final result, with more data in each panel 😉
Cheers,
Mabula
Hi Mabula
The overlap is quite generous at 20%.
Due to the dithering the actual overlapping should be a bit smaller than.
I should experiment with different scale stops.
Registration seems slower with bigger scale stops.
Is it correct?
For Oiii, I had to adjust analyze star settings to include more stars.
Otherwise there was only one panel with incorrect registration.
The settings that I use for Oiii are
Min star size - 4
clip star profile 0.05
filter star profile enabled
kappa - 3
limit # of stars - 5016 (maximum)
With this setting, APP found around 2500 stars per frame.
Hi Mabula
The overlap is quite generous at 20%.
Due to the dithering the actual overlapping should be a bit smaller than.
I should experiment with different scale stops.
Registration seems slower with bigger scale stops.
Is it correct?
Yes, that's correct. If you increase the scale stop, an increasing amount of star patterns are created in both the reference and the frame to register (for pattern recognition) if the amount of stars allows it.
For Oiii, I had to adjust analyze star settings to include more stars.
Otherwise there was only one panel with incorrect registration.
The settings that I use for Oiii are
Min star size - 4
clip star profile 0.05
filter star profile enabled
kappa - 3
limit # of stars - 5016 (maximum)
With this setting, APP found around 2500 stars per frame.
Okay, the amount of overlap of about 20% is really nice, this ensures that registration should work really well and that the data normalization will also be of very good quality.
If the overlap % drops below 15% or even 10% than problems usually will start to occur depending on the amount of stars and nebulosity in the overlap regions. So if you are planning a mosaic, I would advice to have 15% overlap at least to prevent problems in the mosaic creation 😉
Kind regards,
Mabula