Color aberration at star center(s)  

  RSS

(@jbrazio)
Brown Dwarf Customer
Joined: 4 months  ago
Posts: 8
October 16, 2018 19:24  

I'm very new to APP, I've just tested my first integration (HA, S2, O3) and I see the following artifact on some of my stars (zoomed image): 

Untitled

i.imgur.com/PL8wx0S.png

Could anyone point me out on the right direction ?

Thanks.

This topic was modified 4 weeks  ago 4 times by João B.

ReplyQuote
(@vincent-mod)
Neutron Star Moderator
Joined: 1 year  ago
Posts: 158
October 16, 2018 23:17  

Links get removed here most of the time, you can however just add the picture by clicking on "attach files" on the bottom left of the reply section.


ReplyQuote
(@jbrazio)
Brown Dwarf Customer
Joined: 4 months  ago
Posts: 8
October 16, 2018 23:47  

I don't see any "attach file" button.. is it available for all users ?


ReplyQuote
(@vincent-mod)
Neutron Star Moderator
Joined: 1 year  ago
Posts: 158
October 17, 2018 00:17  

Good question. 🙂 It might not then, you can split the link in 2 parts..


ReplyQuote
(@jbrazio)
Brown Dwarf Customer
Joined: 4 months  ago
Posts: 8
October 17, 2018 00:44  

I now see the "attach files" button, did you change something ? If not then user with less than x posts are not allowed to attach files and I've crossed that threshold. 🙂

Untitled


ReplyQuote
(@vincent-mod)
Neutron Star Moderator
Joined: 1 year  ago
Posts: 158
October 17, 2018 02:44  

I actually think that is the case indeed. 😉 It's a safety feature against spamming.

So I now see the issue, I also see some random red dots. Did you integrate using outlier rejection?


ReplyQuote
(@jbrazio)
Brown Dwarf Customer
Joined: 4 months  ago
Posts: 8
October 17, 2018 09:13  

I was not sure, but I went back and redo the integration.

Integration settings: quality, average, 1st deg LNC (3 iterations), MBB 5%, winsor clip k3 i1 d0, lanczos-3 no overshoot, topHatKernel size 1.0, no drizzle scale 1.0.

Files used:

  • L100: St-avg-5760.0s-WC_1_3.0_none-x_1.0_LZ3-NS-full-qua-add-sc_BWMV_nor-AAD-RL-MBB5-Lum_1stLNC_it2.fits
  • R100: St-avg-4800.0s-WC_1_3.0_none-x_1.0_LZ3-NS-full-qua-add-sc_BWMV_nor-AAD-RL-MBB5-Ha_1stLNC_it2.fits
  • G100: St-avg-4800.0s-WC_1_3.0_none-x_1.0_LZ3-NS-full-qua-add-sc_BWMV_nor-AAD-RL-MBB5-OIII_1stLNC_it1.fits
  • B100: St-avg-4800.0s-WC_1_3.0_none-x_1.0_LZ3-NS-full-qua-add-sc_BWMV_nor-AAD-RL-MBB5-SII_1stLNC_it2.fits

Result (zoom ~175%):

Untitled


ReplyQuote
(@minusman)
Red Giant Customer
Joined: 1 year  ago
Posts: 93
October 17, 2018 22:14  

Try activating diffraction protection with a value of 10% of your frame count. And weights Star Shape or Quality. Maybe that helps.


ReplyQuote
(@vincent-mod)
Neutron Star Moderator
Joined: 1 year  ago
Posts: 158
October 18, 2018 00:14  

That might be worth a shot yes, otherwise I have to take a look at the data myself as it requires a bit of playing around I think.


ReplyQuote
(@jbrazio)
Brown Dwarf Customer
Joined: 4 months  ago
Posts: 8
October 18, 2018 00:29  

I'm integration in total 83 images, so I set the diffraction protection to 8px.. it had no effect on the star centers.

Playing around I found that it has somehow to do with the Lum channel, when combining RGBL if I set the Lum channel over 40% I get the effect at bright star centers, increasing up to 100% makes the star center hue shift from yellow to the ones you see on the previous pictures. Does this help identify the problem ?

I may post the link to the full batch if required (2.6GB lights, 4.2GB flats and 6.0GB darks+bias).


ReplyQuote
(@vincent-mod)
Neutron Star Moderator
Joined: 1 year  ago
Posts: 158
October 18, 2018 03:47  

You can also mail the link to vincent@astropixelprocessor.com 


ReplyQuote
(@jbrazio)
Brown Dwarf Customer
Joined: 4 months  ago
Posts: 8
October 18, 2018 13:39  

Hi Vincent,

I've uploaded the dataset and sent you by mail the links.


ReplyQuote
(@vincent-mod)
Neutron Star Moderator
Joined: 1 year  ago
Posts: 158
October 21, 2018 01:09  

So, I followed up a bit by mail. The result I got I attached here (I think I mixed the colors wrong, but that's not the issue 😉 ) and while I see the effect very clearly after combining the data (which is looking great seperately) I can reduce it substantially with a lower addition of the Luminance overall. Also my steps may have caused it to be less pronounced, so here they are;

- Loaded all the data using "multi filter processing" enabled.
- Registered with "dynamic distortion correction" enabled.
- Integrated using the "Quality" weights, stacking 95% of lights, 1st degree LNC/3 iterations, MBB enabled at 5%, MAD winsor clipping at standard values, diffraction protection at 10.

I then combined with 75% Luminance added to the rest at 100% R, G, B (with Ha to red etc). In this case reducing the luminance had a big effect in reducing the star center anomalies (which I guess are due to clipping, although I thought a Luminance addition wouldn't cause that).

To answer a few of your other questions;

What exactly does the “x” and “bg” sliders do ?

The "x" is the factor at which the data is mixed, kind of the weight of that channel overall. The bg is like a stretch I believe upping or lowering the background level which in cases of nebulae in a certain channel can be nice to do, although it didn't have a huge effect here.

All the data seems reasonable and like it should look, in terms of flats, bpm and darks ?

The flats are not great I think, you do see the dust bunnies and it doesn't seem they are of great, bad, influence, but it might be an idea to reprocess without flats and see how that looks after light pollution correction etc. If flats are not ok, it's better not to use them. I think you might need to have different exposures for the flats and also a more even light-source as I'm seeing some weird cut-offs and such. We can talk about the flats in a new topic if you want.

BPM looks ok, bit on the bright side maybe, master dark seems ok but you do see light bleeding at the edges, same for the master bias. It might be that your setup isn't completely dark when you take these fames.

combine RGB image lpc cbg St

This post was modified 4 weeks  ago by Vincent Groenewold - Moderator

ReplyQuote
(@jbrazio)
Brown Dwarf Customer
Joined: 4 months  ago
Posts: 8
October 21, 2018 12:40  

Thanks for taking time to look at this.

I’m far from being an expert on this but I also do believe it has to do with clipping when integrating the luminance channel.

1) The defect scales with the amount of integration of Lum (I was able to integrate 40%, you went to 75%).

2) The defect appears first on the brighter, more saturated, stars.

Why aren’t more people having this issue ? Are my individual channels overexposed ? Is the Lum channel overexposed ?

I really hope @mabula-admin has time to have a look at this, if required I send him the same data I sent you.

Without going to much off topic while we wait for  Mabula to kick in..

I do believe my darks were taken during the night, I thought what I see on the four corners are what people call amp glow, is it not ?

For the flat it’s clear I need to clean the whole imaging train and I even need to check if my filter wheel is working properly as the illumination is rather even I don’t know why so many differences between the shadows on each channel.

I created the flats using SGP 30000 ADU tarde value, when you say the BPM is in the bright side it means I have to lower the ADU ?


Share: