Ok, here is my first attempt at narrow band colour processing which I offer for critique. Object is IC2872_Gum39 in Ha/OIII/SII. Not quite up to the standard set by MnT in a recent post on this image but I'm ok enough with this one to put it out there. Many thanks to Rick S for providing guidance in response to my post on narrow band processing.
About 18 hours of data all up. I used the Ha channel as luminance and fiddled around with dozens of palette combination in PI pixelmath. I had to stop doing it as it was driving me nutty .
Equipment is a TSA120 OTA, QSI683 and Astronomik 12nm filters.
I notice it's very red, and not so much blue. What colour mapping did you use? I'm interested because I know that there's not so much OIII around, and I found it hard to bring out without revealing every evil artifact and flaw in my system.
Your stars look lovely. Small, tight, not bloated, no nasty halos.
I notice it's very red, and not so much blue. What colour mapping did you use? I'm interested because I know that there's not so much OIII around, and I found it hard to bring out without revealing every evil artifact and flaw in my system.
Your stars look lovely. Small, tight, not bloated, no nasty halos.
Great work.
Best,
Mike
Thanks Mike, I did struggle with the getting the right mix particularly given that there is little OIII to be had. I personally don't lean towards the hubble palette as I found that it made too much green for this image. I played around with variations of the colour palettes posted by Juan on the PI site and came up with one that I thought sort of looked ok. I finished adding that to a straight Ha/SII/OIII mix.Attached is a screen shot of the pixelmath setting that I used. It still came out quite red when I stretched it though.
I am aware that the representation of colours attributable to OIII and SII do not exactly leap out but I couldn't figure out how to promote these much more. You have done a much better job in your Chick and the Bird Bath image.
Thanks for the feedback, its good to have other eyeballs reporting. Helps with the learning process.
For interest as well, I have attached screen shots of the master channels.
Good start Rodney, lovely round sharp stars and it's somewhat reminiscent of the CFH colour pallate w.cfht.hawaii.edu/en/gallery/
Theres a fair bit of O3 in your master but imo the S2 is probably lacking in depth, so when combined with that colour palatte, the image is a bit mono tonal and lacks the 3D look in Mike's and other images of this object.
Probably just me but I think that hubble palatte works better as a base for this DSO in NB, but that also needs HLVG or selective colour to reduce the dominant green from the Ha.
It might also be worthwhile reducing your star sizes as well if you can, as they are somewhat distracting atm.
Good stuff, welcome to the dark side of anything goes in NB
Last edited by Andy01; 08-05-2016 at 10:22 PM.
Reason: typo
Nice job, Rodney. It does look quite a lot like a RGB image. Nice to go after one of the less common targets as well.
Hubble palette is nearly always too green to start with because Ha is dominant in most objects. As Andy suggested, there are many ways to dial back the green after the initial colour combine.
looking very nice Rodney, always good experimenting with palettes.
I know it isn't PI but here is a good guide for getting colour out of a green NB image http://bf-astro.com/hubblep.htm
I wrote the ColorMask script in response to a challenge to implement Bob's NB colour mangling process in PI. ColorMask + Curves can do some very fancy stuff.
I wrote the ColorMask script in response to a challenge to implement Bob's NB colour mangling process in PI. ColorMask + Curves can do some very fancy stuff.
I'll take your word for it Rick! That was half jibberish to a non PI user
still a photoshop user might get something out of the link.
I'm sure i'll give PI a go at some stage - its probably nearly worth it for DBE ... perhaps if the weather ever clears up!
Rodney was kind enough to share his data with me to do a quick & dirty Hubblesque process in PixInsight. The result is here: http://www.astrobin.com/full/248149/None/
The processing steps were...
Create luminance:
Noise weighted integration of Ha/Oiii/Sii masters (no rejection)
Light linear masked noise reduction with MultiscaleLinearTransform
HistogramTransformation stretch
Apply star mask: slight MorphologicalTransformation Erosion to reduce stars
Apply HDR mask (clipped luminance with stars removed): LocalHistogramEqualization (add a little contrast)
Green ColorMask: Curves Green reduction, Red boost
Cyan ColorMask: Curves Blue boost
Yellow ColorMask: Curves Red boost
Curves Hue - map Cyan to Blue
Magenta ColorMask: Curves Red reduction
Final image:
LRGBCombine Luminance and Colour
Clipped luminance mask: Curves a* tweak
DaskStructureEnhance script
I probably could have done the colour mangling a little more simply with more time and effort but I did what I normally do and just tweaked away until I was mostly happy.
Happy to explain the steps in more detail if anybody is interested, perhaps in another thread rather than polluting this one too much. This is pretty similar to my usual workflow except that normally I'd remove the stars before doing the colour combine. It's a bit painful but avoids magenta stars and lets you beat up the colour data very heavily to produce nice rich colours.
Thats good Rick and thanks for sharing the processing steps. The key achievement here is to show a better delineation between the different narrowband regions which my version did not. Obviously in this subjective world we can have different views about colour. It is good though to get the most out of our data whatever the palette is that we choose. The only thing I would probably look at in further processing is to push the colour saturation somewhat. Thanks again.
Thats good Rick and thanks for sharing the processing steps. The key achievement here is to show a better delineation between the different narrowband regions which my version did not. Obviously in this subjective world we can have different views about colour. It is good though to get the most out of our data whatever the palette is that we choose. The only thing I would probably look at in further processing is to push the colour saturation somewhat. Thanks again.
Thanks for the loan of the data, Rodney. Hopefully you'll get at least a couple of new ideas from my approach. I didn't push it too far but you certainly could do a lot more with contrast and saturation to make it pop.
Rodney was kind enough to share his data with me to do a quick & dirty Hubblesque process in PixInsight. The result is here: http://www.astrobin.com/full/248149/None/
That's more like it! Nice work gents, thanks for sharing!
Thanks for the follow up comments. Agreeded that this has been a fun and informative thread. I look forward to the opportunity to apply some of the techniques when time permits.
.
