Weird results with ...
 
Share:
Notifications
Clear all

Weird results with Combine RGB  

  RSS

(@rickwayne)
Main Sequence Star Customer
Joined: 2 years ago
Posts: 31
May 9, 2021 03:41  

I did a quick LRGB of the Leo Triplet last night -- bog-standard, every frame was the same gain and exposure, 15 minutes L and 10 each R, G, and B -- and when I brought the integrations into the Combine RGB tool, using LRGB 1, the galaxies glowed this bizarre blue-green color. Looked to me as if the filter tag had correctly set the channel for the images during the import dialogs (I just selected all 4 integrations in the file chooser).

I saved the channels as TIFs and pulled them into Photoshop, copying and pasting Luminance into Alpha, R->R, and so on, and the color balance was perfect.

Any ideas? Maybe the FILTER tag got messed up in the integrations? Since I've got a good image (at least within the limitations of my data), I'm not hurting to solve this -- just wondering if anyone else has encountered a like problem.


ReplyQuote
(@rickwayne)
Main Sequence Star Customer
Joined: 2 years ago
Posts: 31
May 9, 2021 03:46  

Just went back and checked. The FILT-1 tag was definitely correct in each one, though the blue and green channels were completely whited out in FITS Preview (the other two also looked way "overexposed" in FITS Preview but some detail was visible).


ReplyQuote
(@vincent-mod)
Universe Admin
Joined: 4 years ago
Posts: 3392
May 9, 2021 19:52  

Interesting, it's probably faster if you could upload the integrations of L, R, G and B (so those separately) to our server for us to check;

Go to https://upload.astropixelprocessor.com and use upload1 as username and upload1 as password.

Create a directory named “rickwayne-rgbCombineIssue” and upload in there. Thank you!


ReplyQuote
(@rickwayne)
Main Sequence Star Customer
Joined: 2 years ago
Posts: 31
May 9, 2021 20:14  

Thanks Vincent! Uploading now. Dunno if it matters, but Luminance and at least one of the other channels wound up getting processed without flats; I took the flats the next day and obviously the optical train got disturbed, there were awful artifacts particularly visible in the L channel. Everything else was bog-standard, I just clicked the "Analyze stars" button, accepting all defaults along the way, eliminated a few low-quality subs, then "Integrate", with all defaults.


ReplyQuote
(@vincent-mod)
Universe Admin
Joined: 4 years ago
Posts: 3392
May 9, 2021 22:33  

Thanks Rick, will check tomorrow. I need a good nap from a busy mother's-day. 🙂


ReplyQuote
(@vincent-mod)
Universe Admin
Joined: 4 years ago
Posts: 3392
May 10, 2021 11:54  

So I had a look and indeed it is not giving nice RGB colors. Not sure what is going on, would you mind uploading like 10 subs of each filter (including some raw calibration data) as well of the raw data?

Thanks!


ReplyQuote
(@rickwayne)
Main Sequence Star Customer
Joined: 2 years ago
Posts: 31
May 10, 2021 16:23  

Uploaded. The luminance images are in the main folder -- found out that the upload1 account doesn't have move/delete privileges in DSM, which is perfectly appropriate. R, G, and B are in their appropriate folders.


ReplyQuote
(@vincent-mod)
Universe Admin
Joined: 4 years ago
Posts: 3392
May 10, 2021 21:26  

Thanks! I am noticing a bit of a low offset in the luminance for instance, is 8 a normal value for this sensor? This will also be important for the calibration data, you don't want to have too low of a setting for the darks for instance. Will depend on the sensor though and I don't know this one well. I'm wondering if that is responsible for the lack of data in the background like I see in the histogram;

Screenshot 2021 05 10 212333

The Green and Blue lights I don't see in the upload, so if you could still add those and your master dark, master flat and master bias that would be great! 🙂

edit: Also try yourself to integrate your data without the use of any calibration data, to see if that helps. But for sure you need likely a higher offset and a longer exposure. This may cause issues with calibration.


ReplyQuote
(@rickwayne)
Main Sequence Star Customer
Joined: 2 years ago
Posts: 31
May 11, 2021 15:56  

Huh.  Sorry about the missing lights, I swear I dragged them over to the folder but maybe I closed the tab or my Mac went to sleep before they finished uploading.

I was setting the bias based on Cloudy Nights discussions around testing the camera -- ZWO have their own setting scale that doesn't correspond to actual values, or so I read. With only 4096 values to play with, I was concerned about blowing out stars so I kept it pretty low, but if I'm clipping the low end I obviously need to increase it. Wonder why I never noticed that before?

I'll try the no-calibrations integration.


ReplyQuote
(@vincent-mod)
Universe Admin
Joined: 4 years ago
Posts: 3392
May 14, 2021 09:09  

Hey Rick, sorry for the delay I'm downloading the new lights now. Did you have any luck with not using calibration data?


ReplyQuote
(@vincent-mod)
Universe Admin
Joined: 4 years ago
Posts: 3392
May 14, 2021 11:11  

Ok, so I processed without calibration data and did a RGBCombine. I already noticed that not using the calibration data, resulted in APP filling in the missing data nicely on the background. The color signal is very low, which is consistent with the problem of clipping on the low side in the data overall. Turning up the saturation a lot, you do get some nice colors popping up, so the reason this bizarre color was there was because the calibration data was clipped as well, causing issues in the calibration overall. So, definitely raise the offset and expose for longer and you will get some very nice results I'm sure. 🙂

Screenshot 2021 05 14 110752

ReplyQuote
Share: