Vignette Overcorrec...
 
Share:
Notifications
Clear all

Vignette Overcorrecting


(@nharvie)
Molecular Cloud Customer
Joined: 4 years ago
Posts: 5
Topic starter  

I'm struggling with integrations having overcorrected vignette and would like some assistance with troubleshooting. 

  • I am imaging with an ASI533MC Pro on an 8" F5 Newtonian with a MPCC Mkiii and 2" Optolong UVIR filter.
  • I'm in a Bortle 5 area.
  • I sequence using N.I.N.A. 1.10HF3.
  • Flats are created using an A3 LED panel with a white film cover on the scope. I dim the panel to achieve an exposure of at least 2 seconds.
  • Flats are captured using N.I.N.A. Flatwizard with target exposure >2 seconds and <10 seconds and have tested 30%, 50% and 70% ADU.
  • I've also tested capturing FLAT/DARKFLAT using Sharpcap 4 also at approx 30%, 50% and 70% histrogram peaks.
  • All at images at Gain 101 and default offset.
  • Processing using APP 1.083-beta2.
  • For each processing run all data/memory is cleared from load screen. Each integration is saved with its masters in its own subfolder to avoid previous master files being detected used.
  • Integration tests are using default values using 5x each of LIGHT/DARK/FLAT/DARKFLAT.

I haven't adjusted my optical train since last image acquisition and can recapture any flats as suggested, I'm happy to work through this with anyone that has time to help and provide whatever data and test any suggestions. 

Rather than dump a sea of images and logs etc, I'll wait to hear what data and information will be useful for others to assist.


ReplyQuote
(@vincent-mod)
Quasar Admin
Joined: 5 years ago
Posts: 4727
 

Overcorrection can be a thing when the flats have some issue, not being the same illumination profile as the lights somehow. Or maybe a calibration issue. Maybe I should look at your masters and a few lights to try and diagnose the issue.

Go to https://upload.astropixelprocessor.com and for the username and password, use: upload

Create a directory named “nharvie-flatsOvercorrection” and upload in there. Thank you!


ReplyQuote
(@nharvie)
Molecular Cloud Customer
Joined: 4 years ago
Posts: 5
Topic starter  

Thanks for your help! I've uploaded a set of images to the folder, each subfolder has 5x images for the lights and each calibration frame. The flats and darkflats I've uploaded were created with NINA's wizard with a target of 50% ADU. 

I've added this data, including flats created at 30% and 70% of ADU, to a google drive share in case anyone else wanted to see it. I'll leave that data in place until after this issue is resolved.

I hope it's something wrong with the data as that means its something I can control.

For anyone else reading this, my integrations look like :

image

Cheers.


ReplyQuote
(@vincent-mod)
Quasar Admin
Joined: 5 years ago
Posts: 4727
 

Having a look, thanks for uploading.

Doesn't look too bad overall, but the flats are indeed overcorrecting. Already when I just check a single sub. To me it looks like the flat just isn't good compared to the light. I wonder if the panel you're using is maybe not very even?


ReplyQuote
(@vincent-mod)
Quasar Admin
Joined: 5 years ago
Posts: 4727
 

Not using the masterflat, I could still correct it quite nicely with the light pollution tool. Bit of color correction;

test RGB session 1 St

ReplyQuote
(@nharvie)
Molecular Cloud Customer
Joined: 4 years ago
Posts: 5
Topic starter  

Thanks Vincent, you managed to get a good result.

I've re-taken another set of flats/dark flats and paid careful attention to having the panel sitting perpendicular to the scope. I've uploaded these to the folder with V2 suffix.
Unfortunately I obtained the same overcorrected result when stacking with these.  Whilst a good integration can be obtained using the tools as you've shown above, I hope to work on some mosaics in future and would benefit from an automated calibration.

You mentioned the flats didn't look good compared to the light - can you let me know what makes a flat better than another? As I take my flats indoors I can play with many variables to try and obtain good flats.

This is how I've just taken the last set. A sheet of 100gsm paper is glued to a collar which covers the scope. The A3 sized panel is about 60cm away. 

image

If I need to change my technique I'm open to suggestions.

This is my integration result before any further processing:

image

Cheers.


ReplyQuote
(@vincent-mod)
Quasar Admin
Joined: 5 years ago
Posts: 4727
 

That's a very nice setup! Well, I'm not 100% sure the flat is wrong, but I think it doesn't quite match the light because of the over-correction. I'm not seeing big issues with your darkflats or darks either. One thing I notice about your setup is that the panel is placed further away, normally I see people place it directly in front of the scope (I do too). Also without extra sheets, if the panel is good and you can change the brightness on it, it may be interesting to see what happens without extra sheets and close to the scope.

Is the A3 panel full spectrum?


ReplyQuote
(@vincent-mod)
Quasar Admin
Joined: 5 years ago
Posts: 4727
 

Another thing I notice is that the darks are pretty old. This can also cause issues sometimes. Could you try making new ones to check as well? Same for dark-flats etc.


ReplyQuote
(@nharvie)
Molecular Cloud Customer
Joined: 4 years ago
Posts: 5
Topic starter  

Thanks Vincent, I have had some success!

I'd recently used the panel on the scope while imaging and had moved to an indoor setup as I thought my flat issues were related to that. Nonetheless, I took a fresh set of flats as suggested with panel on the scope and no extra covering, this made for very short exposures - something I'd been avoiding as I'd read they can cause issues. 

New dark flats and darks were also obtained, so all new calibration frames using the original 5x light frames.

I got this: woohoo!

image

As so many variables had changed for the above test I was curious as to what calibration frames made the difference. 

I re-ran integration using only the new darks. The lights, flats and dark flats were from the original set I'd been using for testing.

I got this:

image

This is only one test, however for now it looks like new darks has solved the issue for this set of test data. 

I will update my dark library and reprocess some of my recent images to confirm if the issue has be resolved for me on a broader scale.

Thanks for your time and prompt attention to my issue.

In coming days I will update this topic with the outcome of my tests processing other data using fresh darks.

 

This post was modified 6 months ago by Noel Harvie

ReplyQuote
(@vincent-mod)
Quasar Admin
Joined: 5 years ago
Posts: 4727
 

Excellent! It can be quite challenging sometimes to find these issues. So I guess you can't adjust the brightness on the panel, if that's the case, extra layers are advised. Indeed to get an exposure of around 1-2 s. The dark being the cause makes sense, it's very likely that the noise profile of your sensor has changed a bit, either due to temperature or just because it's a bit older again. APP assumes the calibration files are correct of course and probably substracts too much/less noise from the lights or something like that as it's now not representing your sensor correctly anymore.


ReplyQuote
Share: