IC434 - Horsehead Nebula

[jase]

Hi All,
I was going to post this a few days ago, but thought I'd provide some lead time for Martin's image to get the recognition it deserves - very well done. While I've targeted this area before (http://cosmicphotos.com/gallery/image.php?fld_image_id=128&fld_album_id=11), not at such a long focal length. Here's a collaborative effort between Alvin Jeng and I on an object that appears to always be a crowd pleaser and flavour of the month around here...the infamous IC434 - Horsehead Nebula (http://cosmicphotos.com/gallery/image.php?fld_image_id=170&fld_album_id=11)

The Horsehead nebula (also known as Barnard 33/IC434) is a tower of dark dust and gas which harbours the embryonic stages of young stars. The nebula is surrounded by energised hydrogen which glows vibrantly red. The Horsehead nebula resides approximately 1600 light years away.

The image is an LRGB composite totalling 10.3 hrs (L:170min;R,G,B:150min respectively). Alvin acquired the data on Lightbuckets 20" RCOS in Mayhill, NM. Total data collected was considerably more, but many subs where thrown away due to poor seeing and considerably nasty glare from Alnitak (see attached sub to view the dilemma). The glare streaks were most prevalent in the blue and luminance data. I wasn't too worried about the colour data as you can really blur this to remove these type of anomalies, but the luminance was a different situation. To reduce the impact, I blended the red channel (which was barely impacted by the glare) in to the luminance, then followed up carefully with the clone tool. Still nasty and time consuming, but minimised the impact somewhat. Unfortunately, dithering and data rejection didn't help very much in this situation. No Ha data for this target, just luminance. The Ha may have accentuated the curtain like streamers coming out of the background neb a little more, but I think I would have struggled with the RGB blend.

Slowly getting the hang of CCDStack and its functions...nice tool, but I've got some questions from those that use it...
When you perform a heavy deconvolution, I note that CCDStack draws a border around the image (probably around 15 or so pixel from the edge of the full frame). I can understand why it does this as any image misalignment could result in erroneous data for the algorithm to calculate. Hence the reason why cropping is important when using these types of functions. But what if you've already pre-cropped the image...is there a way you can tell the function to use the entire image? I end up with a dark line (maybe 2 or 3 pixels wide) around the border of the image - how do you get around this, other than bring it into PS and use the healing brush...eeekkk!:eek:

Do you need to normalise the individual R,G,B masters before creating a colour image? (this is probably in the manual, just haven't got that far yet). I'm really struggling to get the right colours out of CCDStack. It may not appear so in this image, but let me tell you, it took me considerable amount work to get the colours to what I thought were ok. The severe glare from Alnitak didn't help. I've never had to do this level of adjustment and tweaking in MaximDL. Perhaps its normal?? Just in case you didn't know, you can renormalise over and over as many times as you want or need. Doesn't alter the data (obviously the weights can change) - this was asked at AIC.

On the RGB combine (aka Color | create menu), do you leave the saturation at 1 or lower it accordingly. I find that stars have a magenta hue to them. I'm certain the G2V colour ratio's are correct. When the saturation is lowered they disappear, but so do the other rich colour tones you've worked hard to extract.:shrug:

Sorry for all the questions, but I sense I'm not getting the maximum out of this tool.

Anyway, it was a bit of fun (correction - a nightmare) analysing and processing, but feel it came together well. All comments welcome, and last but not least, have a safe and joyful Christmas and New Year.
:xmas::D

[renormalised]

Sure that's glare, jase??!!!:eyepop::P

Could be the ol' HH doing warp 9:P:P:D

[Hagar]

Hi Jase and all the best for the silly season. I can understand your frustrations with Alnitaks flare. I had the same problem the other night and didn't realise until the image run was finished. Another lot for the recycle bin.

Your final image has come up very nicely.

[iceman]

Stunning image Jase, the depth is amazing.

[renormalised]

Hey Doug, if you look closely at your piccie, you'll see the ghostly (or is that ghastly??!!) form of the Enterprise going past at high warp:P:D.

That's what stuffed your piccies up:eyepop::P:D

[multiweb]

Hi Jase, merry Xmas to you also and all the best of health for 2009. What a stunning photo. :eyepop: I thought Martin's was pretty cool but I have to admit you've topped it with flying colors. Details are amazing. :thumbsup:


Ok I can help a little bit with CCD Stack. The deconvolution tool has an algorithm that looks at the pixels around each star and it needs X pixels radius around the area it's processing. As a result it can only process a subset of your entire picture and will leave a X border because the data needed to treat the pixels it's working on doesn't exist. (as per Adam Block tutorials and some correspondence with Stan Moore.)


You do need to normalise your subs prior to run any rejection algorithm in your R, G and B stacks BUT I don't think you need to normalise the masters R,G and B prior to creating the color. To check with Stan Moore and confirm but I vaguely remember him posting somewhere that normalising the masters prior to combining the color would in fact throws your color balance. Don't quote me on this though, double check the CCDWare forums.



Never used the saturation in CCD Stack. I save the final FIT as a 32bits float then as a scaled 16bits TIFF and do the colouring "to taste" in photoshop.

[marc4darkskies]

Very nice Jase! :thumbsup: You did a darn good job getting rid of that glare I must say!!

You beat me to the punch too!! I've just started on this target myself but I only managed 3hrs of Ha before the clouds rolled in and stayed there!!! :mad2: I was going to post what I have so far - but now I guess I'll have to wait! :D:P.

No, DON'T normalise your RGB masters before combining in CCDStack - it will destroy your colour balance. I usually start with the combine factors reflecting the relative QE of my chip at the centre of each filter's passband (ie R=1.45, G=1.14, B=1.0) (assumes equal length exposures). I then tweak the factors a bit to taste of course. Once balance is achieved I set the saturation to around 1.25.

Sorry, no advice on Deconv - never do it in CCDStack.

Cheers, Marcus

[jase]

:lol: Could quite well be. Though I do like your Enterprise conspiracy in Doug's image. Plausible... Cheers.



Hi Doug, thanks! Pleasing to know I'm not the only victim of Alnitak's notorious destruction of image data. The trick is to suppress the flare with some data that it isn't so prevalent. Wide field shots make it harder to work with as you'll end up removing stars in the process. Have a good one!:thumbsup:


Cheers Mike. Pleased you like it! All the best to you and your family. Keep up the great work with these forums. Looking forward to seeing some of your images once you get the DSLR modded. Thanks.:)



Hi Marc, Thanks for the wishes. Hope you have a good one too! Martin's image is still a stunner in my book. The AO-L can work wonders at these focal lengths. None of the lightbuckets imaging rigs have them installed. Theres undoubtedly reasoning behind their decision. Thanks for the info on CCDStack. Will hit the CCDStack forums. I'm still rather green with the tool. Seems to take me ages to process images of late. I've got two more targets that I haven't finished processing from last new moon. May get a chance over the break, but the presented image will be the closer for 2008. Will kick off with something not commonly imaged in 2009 (assuming I can get this darn RGB data set to cooperate!) ...All the best mate.:thumbsup:



Thanks Marcus. Trust me, the flare is going to haunt me for some months to come.:scared: You can still see it (well I can, but this image is imprinted across my eyeballs at the moment - way too much processing).:)
Thank you also for the CCDStack info. Thought I was starting to understand the work flow with the tool, but all this information from members here is making me rethink this. Think its time to RTFM. Have a good one!

=======
Thanks all for your support.

[dcalleja]

Jase
It looks like a Hubble image, really sharp detail in the head and surrounds. I use CCDStack but really am a novice at it so I bought the DVD from Adam Block and I dip into it when I get stuck (often). I get frustrated with the tool though as it runs out of RAM frequently even on a 4GB laptop - unlike some other tools it stores all the images in memory and can really slow down. If you haven;t already done so you can take the memory warning off in SETTINGS - I find it really annoying to have my workflow disrupted every few minutes.

I;ve not done too much so far with deconvolution so good luck

[Martin Pugh]

Hey Jase....great shot and great processing. I would love to have a 20" RC at my command one day!

Without checking, did the FOV of the 20"/STL limit the field? It would have made for an exceptional shot to have included NGC 2023....Adam Block's APOD of IC434/NGC 2023 was taken with a 20" RC I believe (albeit from Mount Lemmon)...and that is truly spectacular.

CCDSTACK - I never use more than 30 iterations when deconvolving and after many, many tries and experimentation, I can fairly safely conclude that the RGB creation/result is below my critical standards.....so I dont use it for that purpose...in fact, even with the new CCDIS plug-in, alignment often fails....and I mean often....you have to zoom and blink to see it, but it remains inferior to something like Maxim.

CCDStack will chew up memory and bog down any normal PC....I stopped using it, fed up with constant VM problems and dreadfully slow processing.....however, I now have a 64bit Vista PC with 8GB of RAM, and its operation is awesome. I can now have 20+ 41Mb images open, and it doesnt care, and I often run two deconvolution routines simultaneously, each with different settings to compare results.

Well done mate
cheers
Martin

[gregbradley]

Love it. Superb processing as usual.

As Martin mentioned it would have been nice to capture that beautiful little blue neb with the bits of pink streamers in it.

As far as CCDstack I use it and have for some time. I like the way it follows the work flow of processing an image. Martin is correct its memory programming is crapola deluxe and Stan would do well to get a copy of Images Plus and see how it handles memory so well.

With regards to normalising the RGB before combining, my experience is it depends on which brand filters you are using.

When I used Astrodons I didn't use it and using it seem to worsen results.

When I use Astronomik or Baader filters I sometimes use it when I get an odd/off colour combine and it corrects it. So for me it depends on the result of the colour combine if I get odd results or if it looks fine.

I have normally used 1:1:1 combines and that worked fine. Again Marcus probably is being more hi-tech than me and it would depend on the camera and the filters on how close they are to a 1:1:1 combine. Astrodon and Baader market on the basis that they are 1:1:1 and they seem to work out that way on a practical basis. Any slight difference is picked up when doing your Photoshoping anyway.

I always normalise luminance (not renormalise if there is such a thing - normalise means to make the bright and dimmer areas into a similar range so sometimes it can make say a dim green sub too bright if you got less exposure time with one colour due to clouds). I don't know if it really is that important though. Probably more so if the subs varied a lot in quality due to clouds or dawn or some such otherwise if they are much the same I wouldn't expect it really does virtually anything.

I don't use deconvolution too much as it oftens gives a vey harsh result. I suppose it is a tool to use lightly and not get too caried away with it. Certainly not the 100 iterations it seems to be set for as default. Perhaps more like 10? Not sure which one works better and under what conditions - the 2 types of deconvolution in the program.

That new align tool in CCDstack though is absolutely a gem. It does such an incredible job and seems to not be slow like Martin points out it has been in the past (plus the standard program's auto align virtually never works except on really small files).

As far as saturation in CCDstack I tend not to use it as I feel there is more control in Photoshop but occassionally I have increased it if the image looked a bit pale.

More importantly save your final image as a TIFF and RAW not scaled. Scaling will cause you to lose some control. Far better result to save as a RAW and then use curves/levels to bring up your image moving the black point as needed to reduce the noise and get your bell shaped histogram.


Also watch CCDstack DDP it seems defaulted to slightly black clip. It may pay to be selective when using it and not use the auto button too much.You'll end up with a histogram not bell shaped and too hard to the left.


Cheers,

Greg.

[jase]

Jeezz, ask and you shall be answered. Really appreciate the info being provided.:thumbsup:



Cheers Dan! I guess like all new things, it takes time to get use to it. CCDStack is no exception. Yes, I've heard of the memory issues, however I've not experienced this...I am running it on a high end vista64 workstation would helps. I may get hold of Adam's DVD to accelerate the process. Thanks again for your comments.:)



Thanks Martin. :) The FOV of the 20"/STL is what is presented (be it reduced in size (but it only slightly cropped). Its rather narrow. Indeed, capturing NGC2023 would have perhaps completed the overall scene, but Alvin decided to frame it differently to accentuate the long curtain type Ha structure behind the HH. Adam's image was taken with a 24" RC. Its a really nice image of the area. I've actually got the raw data and processing tutorial of this image which I picked up at AIC. Pretty cool to play around with. The data actually has a massive reflection going straight through the HH in which he did a remarkable job of resolving.

Thanks for the information on CCDStack. Interesting experience with the new plug-in. I can't say I've had any issues to date, but it is still limited if you trying to match data at vastly different focal lengths. Registar is still a cut above for this type of work. You've got me thinking whether I should go back to MaximDL for certain functions i.e. colour combine etc. Will do some investigation. I'm still in envy of your tight stars...something which can be difficult to deal with due to seeing or general use of broadband filters. Ha always delivers tight stars, so it great to simply use them in the blend process. Thanks again for the info. Seasons greetings!:thumbsup:



Thanks Greg. Excellent information. Renormalise is simply running the normalise function again. Some believed it could or should only be ran once, but this isn't technically correct. As you indicate, it doesn't alter the data, but the weights for each sub compared to each other in the stack. I'm still playing with deconvolution. I do both heavy and light reditions and blend them in giving you greater control over where you want the details. Thanks for the info on DDP black clipping. Never used it for a permanent data altering function as yet, only screen stretches. I like MaximDL's DDP functionality and use to it so will stick with what I know there. Thanks again and seasons greetings!:)

=======
Thanks all. Have a good one!:D

[gregbradley]

With the blazing Alnitak reflections - a couple of thoughts.

This is not dissimilar to the problem with 16803 chipped cameras and filter reflections.

What handled that with the Apogee was a well designed aperture mask. So there is no extraneous light going around the sensor reflecting off any bright metal parts in the CCD chamber that reflect off the flattener lenses. It doesn't appear to have a reflection issue at all now and has less reflections than my STL/Astrodon setup did which were minor but there and occasionally intruded especially in the green subs. Astronomik probably is the best at this and Baader are good as well. Astrodon Gen 2 may or may not be - I have heard conflicting reports so perhaps depends on your camera. With 16803 chipped cameras I'd say it is a risk.

I digress, perhaps he could make an aperture mask - does his RCOS have an aperture mask on the mirror? You see it could be a turned down mirror edge on the mirror causing this to be exaggerated. An aperture mask for the mirror is the solution. I believe late model RCOS's probably have this feature standard but worth checking.

Does this RCOS use a flattener? I had the US$1500 RCOS flattener and the anti reflection coatings looked inadequate and second rate compared to Tak and AP quality. Not much you can do about that except ask him to take the flattener off if he is using one. The difference in image quality is miniscule.

Also I found flocking the interior of my 12.5 inch RCOS helped with this sort of thing. The base carbon fibre tube had a certain sheen to it and was not coated with a black matt absorbent paint like a Tak, TMB or AP scope is.

Also if he mounted a 16803 camera on that scope you would've had the blue neb and the horse in the same FOV - tell your mate to get a new blue camera heheehe.

Greg.

[strongmanmike]

A good solid high-end equipment horsehead that Jase :thumbsup:

When you use the clone tool to remove the flares how do you know what area to use for the cloning to keep the data acurate?

Mike

[JohnG]

Another delightfull image, Jase, what more can I say, truly delightfull :thumbsup:

Cheers

John G

[Ric]

A wonderful image Jase, you have captured some fantatic depth and clarity in this one.

always one of my favourite areas.

[jase]

Greg, the internal reflection I referred to was in Adam's data where he recently got the APOD. Not in my data set. I only had the reflection from Alnitak to contend with - this was bad enough.



Cheers Mike. OK, here's what I did - I created the red filtered master and used this as a reference as it had very little in the way of flare. I then used this to subtract the difference between the other masters that contained the flare. This provided an image which contained the differences between the two. This delta image was then subtracted against each master. The result is was not perfect, thus I followed up with the healing brush. So in a nutshell, by creating a delta reference image I was able to know what areas to keep and what to "clone" or mask out. Thanks again.



Thanks John. Appreciated.:)



Indeed Ric, its always a great area. Takes some work to do it right, but I gratifying when it comes together. Thanks again.



:lol: Get out of the wrong side of bed Fred? Check the composition my friend...its missing the key ingredient to emphasising those "vertical gas lanes" - Ha data. Having reworked and pushed this data for over a month, I'm not fixing anything this time round. Just enjoy it for what its worth.:)

[Peter Ward]

While I don't think it's your best effort, that is an excellent HH.:thumbsup:
(I suspect the slightly enlarged star profiles are due seeing rather than any instrumental effects).



FYI all current model carbon fibre tube RCOS scopes have matt black intertor coated OTA's...which is a moot point on the truss models...but obviously there are differences in +10 year old OTA design to the current iterations.

My 18 month old RC14.25 has vaslty superior machining/baffling and construction to the first model I purchased, and is easily on a par with the likes of Taka/AP.

[Bassnut]

Sorry Jase, that came out all wrong :doh:.

The filament detail in the head is very nice indeed :thumbsup:.

[Craig_L]

Just stunning detail Jase. Has a superb incandescant quality which I have never seen before.