View Full Version here: : First LRGB combination - M42.
30-10-2010, 08:25 PM
I guess the frustration of the weather and daylight saving is starting to show.
This is my first attempt at LRGB combination in Pixinsight. I enjoyed doing this. There is no doubt room for improvement as expert eyes will tell.
Did I take the image of M42. No... It's a Light Bucket run, if you hadn't guessed.
Sometimes I think it may be easier to sell the gear and only do the processing. Over a few years the outlay may not be too different. But that's only half the fun. And it's probably cheating.
L 10 x 120 secs, R, G and B, 8 x 120 seconds. Used the factory calibrated frames. 34 in total - 68 minutes.
30-10-2010, 09:18 PM
Good stuff! Like you, I've only very recently picked the same target with the same lightbuckets scope (LSB002). Very similar exposures too. Awesome!
31-10-2010, 09:36 PM
I feel your pain... Mostly bad weather or mediocre seeing here in Melbourne so far. :mad2: Glad you're keeping yourself occupied and are honing your processing skills though - time well spent once the season kicks off! :thumbsup:
I'd be happy to hang that one on my wall though, just maybe a slight contrast enhancement? (looks a tiny bit hazy on my 'good' processing monitor, but less so on my analog connected monitor).
31-10-2010, 10:00 PM
Thanks Ivo. I spent the day working on this - teeming - but it doesn't match some of the work being posted.
Here's today's final. Anyway always happy to receive feedback. THat core is difficult to control - it's way too bright. Comparing, I guess it's not a lot different to the first one.
01-11-2010, 09:38 AM
The way to deal with that, is to control its brightness separately in a layer - it'll keep the core happy and make your life so much easier!.
Have you looked into masking, layering then blending? Especially M42 is a prime candidate for the 'fake' HDR treatment (e.g. one layer with the core, one layer with the rest, with a slight overlap of the masks where they transition into each other).
02-11-2010, 05:19 PM
Ivo. A few shorter exposures will fix the core - I'll work on that later this month. I'm much happier with the latest. The individual RGB images were combined using a linear technique before balancing the RGB and L fluxes - and seems to have improved the image overall.
02-11-2010, 06:19 PM
Looks really quite good Ivo.
The core can only be handled with a seperate set of short exposures (about a minute?). You compose a completely seperate image out of these (don't worry about the outer noise that will be present) and then blend it into the middle regions in PS. Takes a little fiddling and trial and error. The thing to watch though is that the transition between the core area and the other middle nebula regions isn't obvious, this is the hard bit to get right...and very few Orions get it right in my opinion.
Have a go :thumbsup:
02-11-2010, 06:29 PM
I think the light buckets idea is a good one, it gives access to the sky when otherwise you would not be able, always a nice object to image m42 it does as mike says need to be blended with slightly shorter subs to bring out the core detail.
02-11-2010, 07:17 PM
I used the easy imaging option which isn't that flexible. I'll try the advanced option and get a few more shorter exposures. I guess some extra 2 min subs wouldn't go astray.
It would make sense to take some Ha and substitute for luminance.
03-11-2010, 08:56 AM
Yep, then blend/mask the 2nd sub in as outlined by Mike/me ('real HDR'). Unless you didn't blow out the core in your single sub data of course. In that case you simply 'fake' your shorter sub by adjusting the gamma until detail becomes visible ('fake HDR').
Your 1st image suggested to me there may just be enough data to accomplish the latter, without having to resort to a second shorter sub, but it is hard to judge from a 8-bit JPEG :)
Sort of like this (exaggerated);
03-11-2010, 06:48 PM
I nuked the core on that first image by trying an over zealous pixel math subtraction of a L mask - not the best approach. The latest image is probably the best candidate for this treatment.
A few more shorter L's on order. But on second thoughts I think I need a few more RGB's also, because the core's are even brighter. I'll work on it over time.
03-11-2010, 07:35 PM
Keep us posted - there's lots to learn from work-in-progress threads like these!
03-11-2010, 07:42 PM
Nice image. You certainly nuked the core but that is quite fixable by using your original subs and selecting a not so stretched core and blend it with the already dressed final image. Looks very nice and smooth so will make a nice image overall.
03-11-2010, 10:00 PM
On closer inspection the core on all the images is over exposed. This is a limitation of the easy imaging option. I've ordered a run using full control to take shorter exposures which might help control the dynamic range.
Might be a few weeks before I get the chance to get back to this. I'll see how it goes.
04-11-2010, 03:44 PM
A combination of erosion, mask blurring, pixel math and wavelet sharpening
Sort of better:question: Some additional artifacts...
05-12-2010, 11:34 AM
This is where the image is at. Ivo - you might still be interested.
Basically. Some additional shorter L, R, G and B exposures.
The image subsets were integrated into RGB and L images which were then combined using HDR techniques.
Finally, after several different approaches - I duplicated the final image and reduced the brightness of the clone to reveal the detail. This was added to the main image using Pixel Math. A little Histogram stretch restored luminosity to the rest of the image.
It's an improvement, and it's not perfect.
05-12-2010, 07:38 PM
All ears/eyes! :P
You're definitely rewarded with a much improved core after all that. Any reason why you're just showing a crop? Would be great to see a the whole thing so we can see a proper 'before and after' (e.g. see the progression)!
Are you still on Linux by the way?
05-12-2010, 09:16 PM
It's getting a bit that way, isn't it. Most people would probably have moved on from this, but that's not my way. I'll keep plugging away until I get it right.:lol:
I would appreciate comments about any artifacts that might be present after this processing - the core image seemed more relevant at this stage.
I'm still using Linux - I use Pixinsight for all my processing, unless I need a little refinement in GIMP, with EAW Sharpen or Smart Redux - hard to beat.
06-12-2010, 09:28 AM
A most commendable personal trait in my opinion - the only way to learn stuff and master your trade/art/hobby.
Ok, here's my honest opinion (for what it's worth); it's actually quite good!
The four stars in the trapezium resolve nicely - no overexposure there at all. You could do with a bit of contrast enhancement and sharpening - there's more detail hidden in that core. No compositing artifacts, banding, etc. that I can see in the core.
Your overall image could probably also use contrast enhancement and sharpening - there is more detail to be had in the seemingly 'opaque' areas. The background levels are somewhat high, not aiding contrast and shrouding faint wisps of gas (that's what the Great Orion Nebula is all about!). A bit of local contrast enhancement should take care of that.
Correct me if I'm wrong, but I think you did a star bloat reduction. Most algorithms start producing square stars when taking this too far on small stars - maybe use a slightly less aggressive bloat reduction. Also, as a result of the aggressive bloat reduction, Iota Orionis is a nice bright white point, but now has a grayish halo, fading to blue. IMHO that is somewhat unnatural looking, though I see it all the time in other images.
There's just one thing I don't completely 'trust' in the overall image, and that is the reddish hue/gradients in the top and bottom right corners. I've scrutinized quite a few Orions and I'm not sure that should really be there in an LRGB image (I'm always open to being wrong though).
Great! Hey, would you mind beta testing something for me in a few weeks? ;)
06-12-2010, 07:23 PM
Ivo. I like what you've done with the image. I'll work on that in Pixinsight. Still coming to terms with tools.
Happy to assist with Beta testing - in user mode I guess?
06-12-2010, 09:38 PM
There's a lot more detail in the outer parts if you use an aggressive stretch. I've found a lot of people don't push the faint stuff hard enough. Push hard and see what is there, then try to reveal this and, at the same time, control the the bright bits.
07-12-2010, 03:42 PM
User mode, yep. I'll send you the details once I got the first beta ready. Thanks! :thumbsup:
vBulletin® v3.8.7, Copyright ©2000-2013, vBulletin Solutions, Inc.