Need some help with my HaRGB workflow

[jjosefsen]

Hi,

So I've been trying to get somerthing out of this data for a couple of months now, and I just can't nail it..

I took some Halpha of NGC7635 and I am trying to blend it with the RGB I also took. I can use the NBRGB script while linear and get a decent blend (color wise) but very little of the fine structures of the Ha data is visible.

Then when I try to blend ind the Ha as luminance I get all sorts of artifacts because the stars are much smaller in Ha.

I have also tried to create a "super lum" by extracting the luminance of the RGB and blending it with the Ha master through pixelmath (max(ha,rgb_lum). This deals with the star size problem, but adds alot of noise from the RGB luminance, while the result in the end is better it is not really that great.

 

Finally I have a hard time getting good blends og luminance and RGB, which seems to stem from a mismatch in brightness and contrast. The only way I have managed to get a decent result is if I use the same autostretch on both RGB and Luminance.

But this means I can't use masked stretch or arcsin stretch, which on the individual masters gives significantly better results.. Is there some trick?

 

If anyone wants to play with the data I have created two sets of the Ha master and RGB master in tiff and xisf formats.

The images are registered and cropped to each other so the fov matches, other than that no processing has been done to them.

https://www.dropbox.com/s/ayrvhlcndzljyjp/NGC7635_Xisf.zip?dl=0

https://www.dropbox.com/s/z6kffvioomxao8a/NGC7635_Tif.zip?dl=0

The data is not great, quite bad conditions of high humidity when it was captured and the stars have been blown out a fair bit.

 

I have had a lot of trouble getting to terms with the processing workflow since going mono..! So I would be very appreciative of any help I can get!

[Adreneline]

Hi.

Well I'm not sure what I am competing with here but before I go to great lengths to write down my process is this worse/the same/better than you have achieved so far?

Adrian

[jjosefsen]

21 hours ago, Adreneline said:

Hi.

Well I'm not sure what I am competing with here but before I go to great lengths to write down my process is this worse/the same/better than you have achieved so far?

Adrian

Hi,

This looks like somewhat like what I  get with an NBRGBCombination in PI, but I perhaps a little better actually, I would say a little more detail has made it through.

I would be very interrested to hear, in broadstrokes how you went about it.

[Adreneline]

I took each image into PI and produced a clone which I stretched with STF and then used as an inverted mask; I then used MLT for noise reduction.

I cloned the RGB image and stretched the original with HT and the copy with ArcSinh; the Ha I stretched using HT.

I saved all three as TIF files and opened them in PS; I used Levels to ensure the backgrounds were consistent for all images.

I combined the Ha with the HT stretched Red channel in 'lighten' mode, opacity at about 75% and re-combined into the RGB image.

I then placed the HaRGB image on top of the ArcSinh image and adjusted the opacity to bring through some star colour.

Adrian

[jjosefsen]

3 hours ago, Adreneline said:

I took each image into PI and produced a clone which I stretched with STF and then used as an inverted mask; I then used MLT for noise reduction.

I cloned the RGB image and stretched the original with HT and the copy with ArcSinh; the Ha I stretched using HT.

I saved all three as TIF files and opened them in PS; I used Levels to ensure the backgrounds were consistent for all images.

I combined the Ha with the HT stretched Red channel in 'lighten' mode, opacity at about 75% and re-combined into the RGB image.

I then placed the HaRGB image on top of the ArcSinh image and adjusted the opacity to bring through some star colour.

Adrian

Thank you so much for this. It helps when. Trying to wrap my head around the problems I always seem to have with combining particularly luminance with RGB, very interesting that you did this in PS rather than PI..

I have been toying with APP today and must say I am very impressed! I have no doubt the same results could be had with PI, maybe even better. But it is just so much faster in APP.. I slapped together an HOO image in APP in 20 minutes with almost default settings, and it looks better than the one I did in PI in many hours.. Definitely looking into this some more..

[Adreneline]

I've never had any great success creating Luminance from RGB; the real thing works for me but not synthesised L. I've also found that trying to use Ha for L never gives pleasing results.

PI is great for noise reduction and taking the image non-linear and I like using PixelMath for NB but I do find PS easier to manage colour. I also find Annie's Astro Actions very useful.

I too have just started experimenting with APP and found it very, very good for creating mosaics - even non-planned ones!

The biggest lesson I still have to learn is knowing when to stop!

Good luck.