Problems with integrating Ha/OIII to RGB

That looks very strange indeed, how do the frames look like just before you use them in RGBCombine? So, the individual stacks for R, G, B and narrowband?

Hi Vincent,

thank you for your reply. Would it make sense to upload the 5 files?

The frames look quite normal, and if I combine just Ha/OIII OR R/G/B then the combined pictures look also quite normal (as you can see in the first two pictures in my first post).

I also tried it in APP 1.083.2, but its just the same outcome.

Best regards

Juergen

Hi Juergen,

Sorry for the delay, been very busy. Yes, you can upload the 5 frames first, just to see what happens here and if I can't see what's going on, I'll likely ask for more data. 🙂

Hi Vincent,

no problem, I think you are quite busy with V2.0.0...

Since I was just in front of my PC I already uploaded the files. I think the file names are self explaining. Ha and OIII was done with l-eXtreme, R,G and B with Astronomik L-2 filter, both with my QHY 268C.

Thank you for your support 🙂

Best regards

Juergen

Ok, having a look now. I notice that the RGB itself seems to combine fine, but as soon as I add narrowband data, something is up. I suspect a data issue here, when I load in the OIII by itself just to check, it looks like this:

How did you take this data and how did you load it from the telescope to the computer and into APP? This kind of interference seems a bit like a data transfer issue perhaps. Or are you processing on an external harddrive?

Would be nice if you could upload like 10 subs of the OIII and Ha, together with the darks, bias/flat darks, etc.

Hi Vincent,

strange! I dont see these stripes in my OIII fits which I sent to you (and also not for the Ha). If I only open this file it looks like that:

I will upload some subs of my l-extreme narrowband and the master dark/flat/darkflat.

At the moment I upload the files. Since I had two sessions (with the same flats) I upload ten files from each session (named M16_S1... and M16_S2...)

Best regards

Jürgen

P.S.: as always I took the pictures with kstars on Raspi 4, the pictures are saved localy on the raspi, after the session I download them to the PC. All other process is with APP

You maybe can notice the pattern in the OIII file when zooming in or out, if I zoom in a lot it's less clear, but you can see differences in the background noise that end up causing these patterns. You don't see that at all? Thanks for uploading! edit: I'm also wondering if the bands can be some conversion issue btw, as APP does convert the values a bit, never seen anyone mention that though, but I'll keep that in mind and inform Mabula if that might be the case.

yes, if I zoom in I also see those lines in the separated fits data. i dont see them in the stacked picture befor separating OIII and Ha. Perhaps the separation creates those artefacts? I used drizzle-mode (Bayer- / X-trans) when creating these separated files.

What I dont see are the big stripes you shown in your picture in your post from 10:31.

But despite that artefacts for me it is not clear why I get those crazy colors in trhe RGBCombine. i' curious what you will find out.

Yes, I will investigate that further with your new data as well, but it does point to an issue in the narrowband data, the bands are just an indication something is up. 😉

Just an extra verification; I assume you did use the "extract-Ha" and "extract-OIII" algoritms?

yes thats right

Hi Juergen,

Could you indicate which ones are taken with Ha and OIII, S1 and S2?

I think I figured it out based on the signal. 😉 When I add a masterdark to the Ha frames (S2), I see this message in the console:

12:02:27 - GENERAL IMAGE LOADER: loading frame D:\Moderator\juergenn-RGBCombine\MD-IG_60.0-E300.0s-QHY_CCD_QHY268C-9ee77c4-6252x4176-all_channels-session_1.fits
12:02:28 - 2) CALIBRATE: performing calibration in 32bits normalized floats...
12:02:28 - 2) CALIBRATE: converting data to 32bits normalized floats...
12:02:28 - GENERAL IMAGE LOADER: frame D:\Moderator\juergenn-RGBCombine\MD-IG_60.0-E300.0s-QHY_CCD_QHY268C-9ee77c4-6252x4176-all_channels-session_1.fits was loaded successfully
12:02:28 - 2) CALIBRATE: Adaptive Data Pedestal : raised to: 4.768E-02
12:02:28 - 2) CALIBRATE: Adaptive Data Pedestal : because 2937 pixels are clipping to 0 without it!!!

Which can indicate that either the frames were not exposed long enough to get well above the background, or that the signal in the darks is too high due to an issue there (going to check that). The OIII (S1) signal is even lower, so that will for sure cause issues and I think this is at the basis of the problem you get when combining signals.

Hi Vincent,

both S1 and S2 are taken with the l-extreme (Ha and OIII). So you can extract Ha and OIII from all 20 files.

Regarding Darks: Maybe the dark datas are too old? I took them last October. Perhaps it would be better not to use Darks in that case?

Ahh, sorry I should've seen you used that filter for both. Ok now that is interesting, I do see a difference between S1 and S2 in terms of overall signal, so maybe it has to do with conditions during the nights? You can indeed try to make new darks and maybe expose for longer with that filter. Not using darks in this case may be a good idea, you can also just use the BPM instead of the darks (if you don't have amp-glow, which I think is the case).

S2 was done the night after S1, all the equipment stayed in the observatory. At least moon was different, but all the other things (should) be the same.

Normally I expose 900s with the l-extreme, this time I tried 300s to reduce overexposed stars in the pictures.

Yes, the QHY268C has no amp glow

I think that may explain it then, the clipping of pixels isn't a big issue, so a bit longer of an exposure and new darks might solve your issue. I would try this data without any calibration first and then just add the flats etc to see when the issue starts popping up.

Hi Vincent,

just an update from my side: yesterday night I took pictures of the veil nebula with l-eXtreme. Again I tried to combine Ha and OIII. I extracted like before Ha and OIII WITH drizzling. Result was a red picture. If I turned down all the sliders for the Ha part of the picture I got a weired colored result.

So I tried extracting Ha and OIII WITHOUT drizzling and combined these Ha and OIII. Result: Pretty picture with nice colors and the possibility to tune the amount of each color.

So in my opinion the weired things here have to do with drizzling. My intention to use drizzling was, to get a little bit more resolution. But now I see, that I have to do it without drizzling...

Best regards

Jürgen

Hi Jürgen,

Interesting! That should also be possible, I've asked Mabula to have a look as well here. Please allow for a day or two as he's also quite busy. 😉

No problem. Its not an critical issue for me (since I have a workaround), but would be nice to know the root cause for this behaviour.

Hi @juergenn,

I just had a look at your data and tried to combine the R,G,B, Ha, O3 stacks with the RGB Combine tool. The weird result that you get with the drizzled data has to do with normalization of your data in the tool, especially multiplicative normalization does not work well. Simply set the normalize option to add-scale or none, since you already normalized the 5 stacks yourself 😉

This is what the RGBHOO formula shosw with add-scale normalization:

Which looks fine to me.

I think the normalize in the tool is not working well because of the drizzled data, especially the O3 shows clear drizzle artefacts indicating that you use too small drizzle droplets for the data that you were combining. Drizzle needs a lot of subs and under-sampled data to work nicely.

So I suspect that creating the O3 and Ha stacks using drizzle with larger droplets would work okay as well.

Mabula

Hi @juergenn,

I have made a nice custom RGBHOO formula for your data 😉

Then you want to fix the gradients afterwards witht the Remove Light Pollution Tool:

And then maybe some HSL selective color, to make red more red and Cyan+Blue more Blue 😉

Cheers,

Mabula

Hi Mabula,

thank you for your help. It was the first time I used drizzle, so maybe I have to look in more detaile into it.

And these are really my data? It looks really great! 😀 I will have another try on my M16 data and see what I get out of it. At the moment I put the narrow band filter data aside, so put them together with the RGB is a work for the next weeks around full moon 😉

One additional question: is it necessary to re-stack and produce new Ha and OIII stacks or is it possible to "re-normalize" the existing stacks? As far as I understood I have to restack due to the drizzle problem, right?

Edit: I just saw that there is in the combineRGB tool a possibility to normalize! I didnt realized that until now 🤪