IC4628 in Ha:[Ha+R]:G:B

[jase]

Greetings All,
Well I have to say, astrophotography certainly tests ones patients. While the weather has finally begun to cooperate, I’ve had other technical difficulties which resulted in further frustration. After receiving the custom-made M72 threaded FSQ/STL tube adapter from Precise Parts, I needed to redo the flats frames and re-calibrate FocusMax (as the imaging train had changed in length). The latter is normally a straightforward affair, but I certainly had issues with creating a good focus V-curve. What I thought looked good was the opposite! This resulted in spending two hours collecting RGB data to finish this project only to find that focus was subtly soft. Arrgggg. I could have potentially still used the data, but I didn’t want to compromise my luminance efforts, so I started from scratch. In addition to this, I found that my original estimate of 5min subs was insufficient to balance the strong Ha data. After some fine-tuning, I re-created an improved V-curve model with eleven V-curve runs. My focus is once again sharp (sigh).

So without further ado, I present IC4628 for your viewing pleasure;

IC4628 in HaRGB (http://www.cosmicphotos.com/gallery/nebulas/index.php?path=./&page=0&img=IC4628%20in%20HaRGB.jpg&idx=1)

Quick words on the image and processing (don’t want to bore you now);
Total exposure time is 5 hours – Ha:120min, R:60min, G:60min, B:60min (all 10min subs). Many may recall my previous post of this emission nebula in Ha, well I reused this data as the luminance, however did not use DDP – manually stretch with levels and curves instead. As the title indicates, the final composite is Ha:[Ha+R]:G:B. The Ha data was considerably stronger than I thought and took some work to balance and keep detail. I was tempted to deconvolve the Ha data as it produces sharper stars, but reduces the fainter nebulosity which I felt was more important to maintain considering the size of this nebula.

Dark/Flat/Bias reduced in MaximDL then registered in Registar, and then (no and then) combined (median) in MaximDL (the usual). I saved the Ha and R as tiffs and loaded into PS – layered using lighten mode to create a Ha+R blend. Ha+R, G, B files then brought back into MaximDL for colour weighting. In addition, I created a standard RGB file in MaximDL which I used for selectively restoring colour. Three layers where then brought into PS – Ha, Ha+R:G:B and R:G:B. Ha set to luminosity, RGB selectively integrated (stars and specific nebula features) using the lighten blend mode. Flattened image then used an inverted layer mask to reduce noise in dim areas that lack detail. Finally, duplicated the background layer and created a high pass custom filter layer mask on the duplicate and selectively highlighted some of the details I wanted to bring out (dust lanes, bright features, globs etc). Flattened again and seasoned to taste. That’s about it. Not sure if I’ve seen the end of this. May rework with fresh eyes.

Thanks for looking, hope you enjoy. Despite the frustrations mentioned, I actually enjoyed seeing this come together the way it did. All comments welcome. :)

[Bassnut]

Now thats exceptional work Jase, very nice.

Very "Crisp", especially the cluster in the corner. Guiding and focus is near perfect.

Registar?, not to say that a fine app, but you should be on CCD stack methinks, allows ultimate control, eg seperately data reject (controlable, it shows the data to be rejected) then sum, rather than median stack.

[avandonk]

That is very impressive Jase. Here is exactly the same area taken with the 5DH and 300mm Lens atf/2.8 about this time last year.. You should be able to flicker between your image and this one as they are the same size.

650kB
http://members.optusnet.com.au/~trlee8/IC4628s.jpg

It is obvious there is just a tad more detail in your image.:whistle:

Bert

[Ric]

Woo Hoo, what a beauty Jase, well done.

Cheers

[Dr Nick]

Wow! thats awesome! ;)

Remember, Astronomy does not only test ones patients, it also tests their neck and back. Try finding M83 when it is at zenith, looking through your finderscope vertically, it hurts the next day... ;)

[danielsun]

Wow Jase!! That truly is a fantastic shot!!
Well done!:thumbsup:

[glenc]

Excellent image Jase. :thumbsup::thumbsup:

[jase]

Thanks for the comments Fred. Balancing the Ha data was a little more complex than I thought. I had quite a few issues with gradients as well. Need a break from processing this image before I revisit it.

I don’t use Registar for the combine function (even though it has that capability), only registration of subs and on occasion registering luminance to chrominance data. I combine the subs in MaximDL (align set to none). When it comes to image registration, I challenge you to find a better product. The Registar software only does two things (registration and combine), but does them extremely well especially when working with data from two focal lengths as it can scale, rotate, skew images so they align with absolute precision. When working with mosaics, Registar comes into a league of its own.

CCDstack is a good processing tool. I’ve only used the demo a few times. I may make the purchase, but the combination of Registar, MaximDL, CCDSharp and PS works well for me at the moment. It’s more about having the right tool for the job, than a universal Swiss army knife. More importantly is to know when to use the tool.

Sum?!? Do people still use that? I’ve started using Sigma Reject, but have mixed results as I don’t understand the algorithm as yet. I’ve read many papers on the median algorithm and how outlier data is eliminated mathematically etc. I’ll stick with what I know for the time being, but will experiment for Sigma Reject in due time. Thanks again.:)



Thanks Bert. :) The only distinct difference between the images is the resolution. The 5DH with a 300mm lens delivers a 5.64 arcsec/pixel (I think), whereas the FSQ/KAI-11000 is 3.52 arcsec/pixel. So the resolution is compromised for the wider field of view. A difference of 230mm F/L. I was going to buy the 300mm L series and a socalastro STL Canon adapter, but the new FSQ comes with an optional reducer-QE. This drops the native F-ratio down from F/5 to F/3.64 with a focal length of 386mm. Field flatness would be similar assuming the 300mm was stopped down a few notches (but would compromise nebulosity, needing longer exposures). With the KAI-11000 it would deliver a 4.81 arcsec/pixel. I’m not certain if that’s really what I want. If the chip had 7.4 micron pixels instead of 9, delivering 3.95 arcsec/pix then I think I’d jump to it. May need to test this further. The 3.52 arcsec/pixel is a real sweet spot for wide field work.



Cheers Ric :thumbsup:



:lol: Thanks Dr Nick. Can’t say I have experienced any neck or back problems looking through the scope. I have pixels running through my veins and as such don’t own any eyepieces only imaging equipment. I’m not a visual person. I don’t even align the telescope with the eye. Everything is done via the CCD camera via plate solves etc.:)



Thanks Daniel. Hope we get some more clear skies soon. :)



Thanks Glen. :) Also thanks again for those Neb. catalogue links.:thumbsup: I have been search through them and slowly building a list based on FOV/pairings. I’ve discovered some interesting wide field pairings that may turn out well (I wont really know until I scale and match in TheSky). Some are separated too far from each other to fit on a single frame. So I will end up producing mosaics to deliver a similar result. Stay tuned!