Trying to improve my DSO images but don't know how [Deep Sky] Processing techniques · DanyJrt · ... · 9 · 426 · 1

Hi all,

I'm very new to mono imaging and recently managed to finish my first LHaRGB project on M78 with 13 hours of exposure.

While processing it (mainly on PI, a bit on PS at the end), I used some M78 images I found on internet (like this APOD for example) as a benshmark for colors and so on, but the more time I spend on it the more "unnatural" it becomes. I have the feeling I'm looking at a painting, not a photography anymore, but I don't know from what this is coming from. The image as been color calibrated before stretching (using SPCC), so might I have added too much saturation ? (In comparison to the APOD picture, it seems to my eyes that I have as much Ha in the background...) Is the image lacking details ? Do I need a new set of eyes ?

I also posted a thread in CloudyNights to ask for community's feedback but received very little - and nothing negative. I really appreciate the comments like "It's not that bad, keep practicing!" but without knowing what needs to be changed for improvement, practice has very limited use.

It is not the first time I get this feeling of working on a painting rather than a photography and have no clue what I'm doing wrong.

If some of you are interested in the data, I can share it, as well as the processing workflow to reach such a result.

I'm considering taking classes (like those from Shaun @VisibleDark for example), but don't know if I need some in PI or in PS.

Any negative feedback is more than welcome. Please be harsh. 

I agree in most of the comments Andrea has made. And I find unfortunate that too many stars are missing due to strong denoising or star reduction. Also some non-stellar structures are missing, again, probably due to heavy denoising (or due to over-usage of "AI" tools)

The image does look overly smooth.  I have used the Hagen method of adding Ha to RGB several times and found it difficult to use at times.  I've started using the method described in this YouTube video (Crazed Conceptions method, https://www.youtube.com/watch?v=LwuiS8N1A-E).  This method or something very similar has been described in several other videos, but I liked the explanation in this one.  What I like about it is that it can be an iterative method of removing the broadband red contamination from the Ha.  After a while, it becomes fairly easy, at least for me, to pick the sweet spot.  The method also allows one to adjust the amount of the modified Ha to add to the RGB.  I suspect you may have added too much.  

George

Thank you all for your comments, these are exactly what I was looking for.

To answer, some of these comments, allow me to provide you with some of the steps I took to process the image.

I first combined R, G and B, then added Ha, and finally added Lum.

Regarding Ha, I didn't substracted this channel from the R channel, that might be the reason why there's too much of it. Instead, I stretched the RGB image, then stretched the Ha channel and used Bill Blanchan's script to add Ha to RGB. I didn't recall he mentionned to substract Ha from red but I probably missed it.
I will try the Hagen and Crazed Conceptions methods, thank you both for the recommendation!

I used NoiseX four times in the process, once on the RGB image before adding Ha, once on Ha, once on Lum and a last time overall. I might have been heavy handed on the smoothing indeed. The image seemed grainy and noisy, but I probably shouldn't do pixel piping after 100% zoom. 

Regarding stars, I shot only 30 seconds exposures under R, G and B to use it a star layer, so they shouldn't be overexposed. I probably stretched it too much, so thanks for the observation, I'll try that again.
I also agree with Die Launische Diva, I'm surprised how many stars are missing. I didn't use any star reduction method so I guess the strong denoise I mentionned earlier is the cause of their disapearance. 
In the workflow, I didn't use BlurX on the RGB layer, but used NoiseX before using StarX. I should try again with less denoise and some BlurX before extracting the stars, maybe I'll get more stars in the star field. 

You also all mention that is image looks flat and that the background doesn't stand out. 
Due to the framing I chose for this image, we can see that Ha signal is everywhere accross the image. How can I make the background stand out if the Ha signal is all over the place ? Reduce Ha signal to get to see the dark sky in the background (top left corner for example) ? Increasing contrast between highlights and shadows would help to make the image less flat  - so darken the dark structures ?

Finally, regarding exposure time, my current RGB master file is composed by 30 seconds of RGB and 2 minutes of RGB. I blended both in the stacking process to get higher dynamic range. But I guess the exposition per sub is too low, I could probably try to shoot 5 minutes subs for RGB and get a stronger signal (for now, the faint signals have very few levels, I think that might be an indicator that they are under-exposed).

Many thanks again to all of you, these comments are very precious !

andrea tasselli:
My (possibly final) advice to you is to ditch all these "methods" and go for the heart of the matter: if you're shooting for reflection nebulae go for (L)RGB (and loads of it). Ha is just a side serving, not the main course. Have a look at some "natural" (that is, not NB-enhanced) image such as this one of mine (but there should be tons around in AB): https://www.astrobin.com/zcxbzt/?nc=&nce=

That, to me, is what the overall colors and tones should be looking like (but with better depth and resolution in your case).

Thank you for your input Andrea ! 

You're right, I should probably aim for a LRGB image and get it right before considering adding Ha. 

I like the image you posted, did you shoot osc or mono ?