PixInsight Forum (historical)
PixInsight => Tutorials and Processing Examples => Topic started by: Silvercup on 2010 May 02 15:31:09
-
Hi All:
I just wrote my first tutorial: M27. Revealing external faint halo from an underexposed H-Alpha integration.
In this document I describe a processing workflow that focuses on enhance faint details on underexposed images using AtrousWaveletTransformation and PixelMath. We also describe using of StarMask, Deconvolution, noise reduction with ATrousWaveletTransform and MorphologicalTransformation.
For this example we have used an image of M27 acquired by Dave Halliday with a Vixen Visac at f/9. It's an integration of 25 frames of 420 seconds in the H-Alpha emmision with an ST-2000XM 2x binned. The image has an initial Histogram strech. Despite of short subexposures we are able to reveal external faint halo.
Thanks to Dave for letting me use his image.
I hope you found it usefull.
http://www.aacadiz.com/zona_trabajo/tutorials/m27/en/m27procexample.html (http://www.aacadiz.com/zona_trabajo/tutorials/m27/en/m27procexample.html)
Best Silvercup.
P.D. Mañana lo subo en catellano. Un saludo.
-
Very well done. Thanks for doing it.
Steve
-
Well done Silvercup,
This is an excellent tutorial that demonstrates how well PixInsight can extract high quality information from what initially appears to be sub-optimal data (no offence Dave ;)).
It also demonstrates that the extraction of that data is NOT a trivial sequence of steps. Each step needs to be understood and the processes required to achieve that step need to be fine tuned given the data that is being manipulated.
What is definitely shows is that PixInsight CAN be used to extract the information WITHOUT having to resort to 'airbrushing' (sorry to throw this into the discussion Bud, but you can see where Juan is coming from when you see an image being processed this way, as opposed to the 'user-defined' or 'manual' approach currently needed in the likes of PhotoShop).
Of course, what it also MOST DEFINITELY shows, is that Silvercup has rapidly risen to 'PI-Jedi' status, not only for creating such a stunning transformation to Dave's original data, but also for being able to put together such a clear and useful tutorial to help others try to achieve similar results.
Well done indeed.
-
Silvercup,
This looks like a great tutorial.....I have a few hours of time on this object with my QSI and FRC300 that I have never processed.....this will be a great tutorial for learning the included processes......thanks for taking the time to make it.....I'll give it a try in the next few days.
best regards,
bud guinn
http://www.budguinn.com
-
Silvercup
Boy that was some bit of work..!!
I got a bit dizzy trying to understand it all... >:D
The pragmatist in me can't help wondering if just switching to f/6.4 is easier...(!!)
I also wonder how much damage the moon being out does...(plus my bad LP...)
The Goldman image is what I am shooting for..(in HaRGB..)...eventually.
But it seems I will not get there,even with EXPERT processing,if I don't change the imaging plan..
By the way..were you working off the tif stack..?
I forget if I sent it...
Again,many thanks...I have a lot to learn !
Dave
-
Hi
Would like to look but the link seems to be broken :'(
Harry
-
My heart is still bumping. For one moment, I thought somebody had hijacked our website! ;D
Excellent work, Silver. A really nice tutorial. To both you and Dave: Can we use your text and images to include this as an official tutorial on PI's website?
-
Hi All:
Harry our server is down.
Juan, It was more easy to clone PixInsight tutorial than make one from zero. You can add to tutorial sections if you consider that I haven't made any dumb thing. I will be very glad.
Dave:
If you shoot at f/6.4 is easier to guide and capture more light and more field.
The moon should not be an inconvenience to capture in H-Alpha, the moon and this filter works very well, I do not appreciate image gradients. Obviously no moon is much better. However, in OIII moonless shoot is a must.
Don Goldman's image is made with two emission H-Alpha and OIII. I don't know how he mixed to obtain the image in RGB (well, I imagine). What I can see is that the red channel does not correspond exactly with the H-Alpha emission . The image has a lot OIII and therein lies the difference.
Here is a comparison of M27 with different filters:
http://dg-imaging.astrodon.com/gallery/display.cfm?imgID=148 (http://dg-imaging.astrodon.com/gallery/display.cfm?imgID=148)
As for changing the imaging plan, triy to get OIII on a moonless night and longer exposures to 20 minutes, with the new mount you should have no problems. Take three or four more 20 minutes exposures of H-Alpha.
Your image has H-Alpha and is similar to Robert Glender's one at this link:
http://apod.nasa.gov/apod/ap030725.html
Best, Silvercup.
-
My heart is still bumping. For one moment, I thought somebody had hijacked our website!
That said Juan, it is a clever way to avoid having to sit down and actually 'learn' how to put together a web page.
I seem to remember that, not so long ago, 'someone' did something very similar when they were trying to figure out how to put together a PJSR script O:)
Cheers,
-
Hi
Looks excellent ;) Just what people need when they get past my basics ???
I will goo and read it again 8)
Harry
-
Very nice :) thanks for sharing, and your hard work on it!
-
Silvercup/group
I certainly have no objection to any use of the image.
PM me if you need the Tif stack (out of DSS...)
That link to the Goldman site is most excellent..!
I was talking with N Fleming about how he sometimes use just the two strongest NB channels,and uses a "blend" of the two to make up the third channel.
M97 for example is ALL Ha and OIII...
Sort of OT here...but in P math ( >:D) do two images have to be the same format (ie 32 bit tifs,etc) to do stuff..??
My P math is "not working"..when I try blending ("red" + .4Ha...)..
for example...
They are not identical images either (FOV wise)...but I believe they are registered.
Dave
-
Gracias Silver, ese tutorial puede ser muy interesante, para esos halos de los objetos que se intullen pero que cuestan un HUEV... sacarlos sin llenar la toma de artefactos.
Por cierto lo has sacado en castellano, o no lo he sabido ver ^-^
Saludos, Maxi
-
PS
It would appear that most of the outer stuff is OIII... >:D
Based on last nites images...
Again SII is a blank.
Need help using Pixel math the create a third channel out of ? the Ha AND the OIII... :'(
I will post later...
Dave
-
Hola Maxi:
El tutorial tiene como objetivo precisamente eso. Recientemente pude reprocesar una toma de M81-M82 en la que su autor afirmaba que no había ni rastro de IFN y mediante este procedimiento la logre sacar, muy débil, pero ahí estaba y por supuesto todas las estructuras coincidían con otras imagenes ya conocidas.
Aunque el tutorial intenta ser minucioso yo aplicaría tambien otros procedimientos que en el tutorial no he incluido por razones de brevedad, por ejemplo, en el primer paso quizas habría que aplicar una mascara para no eliminar ciertas partes visibles de la nebulosa. Habría que aplicar mascaras de estrellas especificas en cada caso, por ejemplo para la deconvolución, etc.
Lo que si quisiera resaltar es la diferencia entre la resta de las escalas pequeñas que hay en otros tutoriales y la de este tutorial. En otros tutoriales de procesamiento por escala se hace una simple resta, pero esto puede suponer en el caso de que haya estrellas grandes que queden halos en la imagen de gran escala que despues se traducirán en halos en la imagen final. Mediante la formula de PixelMath del tutorial, que es la expresion utilizada comunmente para mezclar una mascara de luminancia con una de estrellas invertidas, evitamos ese problema, sobre todo si se hace bien el resalte de las pequeñas escalas para que enmascaren las estrellas gordas. Nos queda solo el fondo que es lo que nos interesa rescatar.
Por otra parte, en el tutorial se aplica 2 veces PixelMath pero también se podría haber procesado independientemente la imagen de gran escala y despues añadirla a la original como en otros tutoriales.
En cuanto a la traduccion, por motivos de trabajo imprevistos no pude hacerla ayer, a ver si hoy tengo un poco más de tiempo.
Un saludo, Silvercup.
Hi Maxi:
Tutorial main goal is just that. Recently I was able to reprocess an M81-M82 photo in which the author claimed that there was no sign of IFN and by this procedure I was able to show, very weak but there it was, and of course all the structures were consistent with other images already known.
Although the tutorial is intended to be thorough, I would also apply other procedures that in the tutorial I have not included for reasons of brevity, for example, perhaps the first step it should be applied with a mask to avoid removing visible parts of the nebula. Masks should be applied expecificaly in each case, for example Star Mask for the deconvolution, etc.
What I want to highlight is the difference between the subtraction of the small scales that are in other tutorials and PixelMath operation in this tutorial. In other multi-scale processing tutorials we do a simple subtraction, but this procedure, in the event that there were large stars, can produce halos in the large-scale image and then translated into star halos in the final image. Through the tutorial PixelMath formula, which is the expression commonly used to mix a luminance mask with a star mask inverted, we avoid this problem, especially if we done the highlight of the small scales properly to mask the large stars. We obtain only the background that is what we want to rescue.
Moreover, in the tutorial we applies PixelMath 2 times but large-scale image could also have been processed independently and then added to the original as in other tutorials.
As for translation, for work reasons I could't make it yesterday. I'll try today if I have a little more time.
Regards, Silvercup.
-
Hola:
La versión en castellano (Spanish version):
http://www.aacadiz.com/zona_trabajo/tutorials/m27/en/esm27procexample.html (http://www.aacadiz.com/zona_trabajo/tutorials/m27/en/esm27procexample.html)
Un saludo. Silvercup.
-
Excellent work! I was away on vacation for a week and missed the original post somehow.
-
Silvercup
OK...so having the Ha M27,and now some OIII M57...I am KEEN to try this...
BUT...when I try to multiply the images in step 2...I realize I do not know how to use Pixel Math.. :-[ :-[ :-[
How do we multiply...?
How do we "edit"..or "parse" for that matter...??...when I tpye "+" it seems to add,and "_" subtracts...otherwise I seem lost...
Do we somehoe "tell" PM what keystrokes mean what math function..?
How did PM know what *~ means...?
Sorry,but I am lost...
Also lost about what you did with "curves" to the image after the "ATrous" thingy...
Could I have not just stretched the residual image in histo...?
I need REAL simpelton advice,if possible >:D
PS
Could you put up whatever you get from M57 just to keep me motivated here...?
Hoping for more help...
Dave
-
Hi Dave:
You must read about operations on any programing language. Anyway we only use simplest operations usually.
+ is sum, - is substrac, * is multiply, / is division
If you want x raised to y you must use x^y, and so on.
If you click Pixelmath Edit button it opens Pixelmath Expression Editor, in the right pannel you will see Functions, Operators and Punctuators. Under Operators you will see all Pixelmath operators.
(http://farm5.static.flickr.com/4038/4582723748_a777b00943_o.jpg)
One of them is ~ that means invert.
In Step2 you must multiply Original Image (clonned) by the inverted of the Small Scales Image ie the image generated in Step 1. Pixelmath knows what to do ;).
After Atrous I do a CurvesTrans like this aprox.
(http://farm5.static.flickr.com/4028/4582135587_3d3029d078_o.jpg)
See your m97 Image06 on Filckr :)
Best. Silvercup
-
Silvercup
As Bogart said..."this could be the beginning of a beautiful friendship"... >:D
Also appreciate that you are also on the "poor mans website" (Flickr)
I think it just goes to show how lost we can be (in downtown Paris..) until someone just points out the Eiffel tower,and says "walk in this direction..."
Maybe this goes back to R Nulmans issue...(!!)
OK..enough bad metaphors...I will try this math...
thanks a bunch
Dave
-
Excellent work Silvercup, thanks for sharing it ;)
-
Hi:
Thanks for the comments. I'll try to do a Common Tasks with Pixelmath tutorial for new users.
Best. Silvercup.
-
A PM tutorial would definitely be appreciated. Look forward to it.
Steve
-
Hey Silvercup, that's a great idea. You may focus on image-image operations, such as the one needed to create or merge masks. Also a manual "calibration" example would be nice.
-
Here is the latest result,and some details.
Silvercup has been slipping me tips on Flickr...
Thanks !!
http://www.flickr.com/photos/daveh56/4600250038/
Dave >:D
-
How much is your drift per image Dave? Seems your stars aren't very round. I'd try to fix that first. It'll sharpen up your images quite a bit.
-
Sander
Its around 2 pixel/frame...ie over 20 minutes. (as per DSS info)
Focus using NB filters is problematic for me..I am going to try using the Astrodon parfocal clear (NB) filter... :-[
I am also using "Morphology" to shrink my stars...but a ratty coma remains..
Settings in ACDNR remain an issue also...my stars seem to go a bit fuzzy...should try some star masked decon on them...
Any other excuse I have left out...? >:D
I am working the whole set of issues...slowly.
Dave
-
Dave, 2 pixels per image is much too much. My recommendation is to focus on getting that resolved rather than trying software 'tricks' to fix your images in post processing. I don't know if you bought your CGE Pro hoping to improve star shape but most likely it will not until you remove the differential flexure. You need to find and fix the source of differential flexure in your setup. Once you get below 0.5 pixels per sub your images will get dramatically better.
Unfortunately it's hard to debug differential flexure from a distance. I grab imaging and guide cam and try to move them relative to each other. If that doesn't show where the movement occurs I examine each joint between the two cameras. If those are all rigid the optics are the last option. I would think your Vixen has no internal flexure but it could happen. Check the focusers and all adapters, rings etc. first.
-
I forgot to add that I'm glad you at least *know* your DF per frame. Many people are clueless about it and instead blame poor guiding or optics. That's why I wrote that wiki page :) You can't fix a problem till you identify its source.
-
Hi Dave,
In my case, DF is working against me now, limiting my maximum exposure to about 1/2 worm period (on my LX90, that is around 4 to 5 minutes).
Whilst I await you 'throwing out that 'rubbish' CGE Pro', I am giving serious consideration to the TS OAG-9
(see http://www.teleskop-express.de/shop/product_info.php/language/en/info/p641_Off-Axis-Guider---nur-9mm-Baulaenge---T2-Anschluss---fuer-CCD.html (http://www.teleskop-express.de/shop/product_info.php/language/en/info/p641_Off-Axis-Guider---nur-9mm-Baulaenge---T2-Anschluss---fuer-CCD.html))
as a cost-effective solution to at least eliminate the DF component in my imaging setup, allowing me to then concentrate on whatever residual problems I may then have.
Obviously, with the weak dollar-euro rate at the moment, I will gladly send over an OAG-9 in axchange for the CGE Pro, and would even consider covering the postage, you being a good mate and all . . . . . >:D
Seriously though, it might be a 'food for thought' idea for you.
Cheers,
-
Yes, using an OAG will allow you to eliminate DF or at least reduce it to very low levels. Of course the Law of Preservation of Misery dictates that you're merely swapping one set of problems for another. Guiding with an OAG is considerably harder and more restrictive than using a guide scope.
-
Hi Sander,
Having had no experience of using an OAG, can you clarify "considerably harder" and "more restrictive"?
I was seriously toying with the TS OAG-9, and am trying to wade my way through the choice of 8300-based cameras to choose from as well (my 'current' 8300-based camera isn't that great, and it might yet be heading back onto eBay to see if I can recover the $500 I paid for it).
I have just worked my way through the long thread on CN, where both you and Harry posted, where the thread initially addressed the Orion AO unit, but went on to discuss the new 8300-based CCDs that were coming down the line.
Decisions, decisions, ???
Cheers,
-
Hi Niall,
you'll find this is a controversial topic depending on where it gets discussed. Try asking about this in the autoguiding Y! group :)
- you have a smaller field of view and because you're looking at the edges of your field of view you can have more distorted star shapes than a guide scope would present. I ran into this with my C11.
- Getting both the imaging and guide cam to focus at the same time can be challenging depending on the OAG model.
- longer FL numbers mean the field of view of your guide cam is small and less light per star hits it. So you are restricted in two ways:
- fewer stars to choose from
- dimmer stars
both these restrictions mean you may have to rotate your camera to make sure you get a bright enough star on your guide cam. The guide star dictates image framing in that case. Not desirable.
Image train gets more complicated and you may run out of backfocus space. Or the spacing between flattener and sensor is compromised.
Anyway, it all depends on needs and wants. I can get away without an OAG right now. If I have to I would use an OAG but not for fun, only because it's the only available option to get the image quality I desire.
-
Hi
I always use a OA guider as my big old newt bends a lot ???
Personally I have no problem at all using my SX OA , I just shove it on and never have to hunt for a guide star 8)
But very key to this is my use of the Loadstar guider which is very sensitive and it is more difficult with a less sensitive guider , I also have a sxv guide head which
I have tried with the OA unit and this is not so successful as you may need to hunt for guide stars, I highly recommend a good guide camera , makes life so much easier :laugh:
You will have to us know which 8300 camera you choose
Harry
-
Hi Harry,
Right now I am seriously thinking about the Atik 383L+ - because it is fundamentally price in UKGB£, there is already a HUGE price advantage compared to buying a product fundamentally priced in $ (in fact, it is the only camera that I know of whose $price is higher than its £price - and that is a very RARE situation in rip-off Britain :moneyinmouth:)
Cheers,
(and, we could move this to a new thread if others are interested)
-
Hi,
Is this tutorial still available somewhere? I found the text on web archive:
http://web.archive.org/web/20100511074245/http://www.aacadiz.com/zona_trabajo/tutorials/m27/en/m27procexample.html
The pictures are missing though :( So it is pretty hard to understand what is going on. I'd love to see the pictures back on the aacadiz.com website.
Cheers,
milosz
-
Hi everybody,
I read the whole post, and it sounds really exciting!!
Finally i wanted to check out the first link, but it doesn't work, there is a spanish error message, something like server down or so :o
Thanks for the big work even if i not yet could admire it..
-
Hi:
I'll try to re-upload tomorrow.
Silvercup
-
Thx cause i am kind of curious ^-^
-
There is a problem accessing the tutorial still. In my case, when I do connect, my computer security intervenes and won't download anything from that site. Other times, the site just times out.
-Jeff
-
Still the same on my side :(
-
Is your server back up, Silvercup? I have been unable so far to find a working link to your tutorial.
Dave
-
Hello,
I tried to click on the provided link, but my antivirus software blocks opening the page, saying it detects a Trojan horse in it.
Clear skies
Antoine
-
I just checked the link to this tutorial on December 14th (2011), and the link still seems to be broken.
If the tutorial can be re-posted somewhere, that would be great. I'm dying to figure out how to do multiscale processing. I am using RBA's tutorial (on his website) and my notes from his recent SJAA talk, but I'd love to have as many sources of information as possible.
Multiscale processing seems to be a Pixinsight `killer app', but my efforts to `reverse engineer' this type of workflow is proving to be somewhat difficult. Hopefully I'll figure it out eventually, but any available tutorials are very welcome!
-
I tried to click on the provided link, but my antivirus software blocks opening the page, saying it detects a Trojan horse in it.
I had this as well sometime ago. It did not happen with Firefox, and it disappeared when I reinstalled Win7.
Georg
-
Hi,
This link doesn't work for me. Isthe tutorial still available?
Tom.
-
Don't worry folks...
Just got a 30 cm f/4.5 mirror from Owl Optics..
We can start over soon......... :D
-
Hi, I tried to read the tutorial but the web server says "page not found". Could you please fix this?
Thanks
Nicola
-
Hi All,
The link is still down :'(
And it would be very very interesting for me...
-
Would really like to read this, but both the English and Spanish links are broken.
Geoff
-
I agree. I've come across this post several time now and am always disappointed when it fails. Any chance someone save it to their hard drive while it was available?
Craig
-
Hello
I made a personal copy in PDF and a word doccument.
In order to not violate copyrights, is there a way to sare it?
Regards
Geert
-
Hmm, good question. If a work was previously shared freely, does that mean it can shared again as long as proper credit is given? Does anyone know how to get in touch with Silvercup to approve this?
-
depending on his (her?) forum preferences a forum private message might generate an email. s/he also has a flickr account...
-
Link is definitely down....
-
Hi all:
Sorry for my delay, I had lost tutorial, but finally I have found on a external HDD.
I have uploaded the tutorial to my dropbox account.
English version is here:
http://dl.dropbox.com/u/12061910/M27Tutorial.zip (http://dl.dropbox.com/u/12061910/M27Tutorial.zip)
Spanish version is here:
http://dl.dropbox.com/u/12061910/EsM27Tutorial.zip (http://dl.dropbox.com/u/12061910/EsM27Tutorial.zip)
Just double click on index.html
Hope this helps.
Silvercup.
-
Thanks Silvercup I'll have a try ASAP! 8)
Cheers
-
Silvercup, thank you for taking the time to find this tutorial. Definitely one that will help many of us.
Craig
-
Hi,all
Sorry, I dont uderstand how can I load this tutorial.
Is it possible explain it ?
kolec
-
Just click the link, and download ZIP file. Extract content somewhere (choose any folder) and open it.
-
Thank You Tom from Blava
-
thanks for tutorial
I managed to download it
-
Is this tutorial still available anywhere? The dropbox link is failing.
Andrew
-
Yeh, broken Dropbox link and a nice Chinese page on the original link on page 1
:sad:
-
Not to surprising that this dropbox link is dead since it dates back to 2012 and there had not been any activity in this posting for over 5 years.
Mike
-
it is in the wayback machine, however, none of the images were scraped so just the text is left.
rob