PixInsight Forum (historical)
PixInsight => Tutorials and Processing Examples => Topic started by: dayers on 2011 September 07 17:25:56
-
I have had good success following Harry's good video tutorial about adding Ha to a galaxy image. Has anyone tried the same general procedure with an image that is mostly nebulosity, like M8? I can't find a set of PixelMath parameters that doesn't result in black "stars" burnng holes in the nebulosity. Trial and error is giving me no joy.
Maybe trying to add Ha to an image that is mostly Ha to start with doesn't make sense.
Dave
-
I'm using something similar, found here in the forum, again mentioned as Vicent's method:
C = r - ha
Ha_clean = ha - (bandw_Ha/bandw_R)*C
HaR = R*r_mult + Ha_clean * ha_mult
The trick is to produce a good C image without too many holes. A few are ok, the final result will take care of them. I have changed it a bit to match different time in subs and different bin, making some assumptions (that I'm not sure if they are correct, at least worked in one case) :
C = r - (tR / tHa) * ha
where
tR = time R * bin
tHa = time Ha * bin
But still, in some cases I had to click on "rescale result" in order to get reasonable results. The last time I used it, I first "linear fit" R with Ha and then I produced a very good C with just C = r - ha .
You can see some results here (http://www.astrovox.gr/forum/album_pic.php?pic_id=13198), here (http://www.astrovox.gr/forum/album_pic.php?pic_id=12984), and (a very bad one) here (http://www.astrovox.gr/forum/album_pic.php?pic_id=12911). Zoom in Elephant's trunk and notice that the blue (from reflection) on the nebulosity has indeed been preserved, although Ha in this area is very strong.
What I have found is that you still need GOOD RGB frames, otherwise the final result is not very good.
Since I'm novice on this field, this is just my 2 cents.
-
Hi John,
Thanks for your comments. It sounds like PixelMath is our friend in cases of this kind. I need to find the thread about LinearFit again to see how it might be applied. So much to learn and so little time! I've been using Jim Misti's M8 raw data to practice on.
Dave
-
Dave,
I did add Ha to the Medusa Nebula, an image I posted in the gallery a couple of months ago. Here's the link:
http://pixinsight.com/forum/index.php?topic=3223.0
Ha was combined with the Red master and the new HaR image was used as the Lum channel. I used Vicent's method to combine the Ha and R data. The method is beautifully demonstrated in Harry's tutorial.
Steve
-
Hi Steve,
Thanks for your thoughts, and I will try using the HaR as Luminance. It seems there are a lot of ways to skin this particular cat!
Dave
-
HaR as L ? And how about G and B? A lot of people use this trick but destroys the other colors. Take a look on this http://pixinsight.com/forum/index.php?topic=2036.msg13106#msg13106
(although this Medusa is very good !! )
-
I too am struggling with adding Ha to nebula images.
Could someone maybe please put together a brief tutorial on the best way to do this for those of us who are getting started with PixInsight and neophytes to using PixelMath?
Thanks,
Kerry
-
Hi:
Perhaps you can try a HaRGB Combination script that I have developed.
(http://farm7.static.flickr.com/6156/6169031015_f9f92899d4_b.jpg)
(http://farm7.static.flickr.com/6153/6169565372_62ff356f0c_b.jpg)
The script is very simple to use:
In Control Channels we choose the Ha and RGB images, the filters bandwidth and the Ha multiplication factor.
You will realize that lowering the RGB Bandwidth fewer Ha passes down and viceversa, but the differences are not very evident.
The multiplication factor is more important because it will increase more or less the intensity of Ha, if we have a noisy channel we can not multiply much, if there is more signal we can increase multiply factor (in Harry's example up to 4) .
In Color Calibration Control we can choose whether or not to color calibrate the final image. Uncalibrated is selected by default for script to go faster. Checking Color calibration will calibrate the final image, for it, we will choose a preview of the background and a preview (or the entire RGB image) for white balance that you PREVIOUSLY defined in the RGB image. Depending on the image we choose Layers Structure detection or not. Finally, we can apply or not SCNR to green to finish calibration.
Checking AutoSTF we can apply an AutoSTF when we use a linear images, otherwise we would not see anything. In the case of combining with stretched images you must uncheck it.
Show Buttons. Show HaRGB and RGB buttons allow us to see the RGB and the HaRGB combination, both buttons open the console while working, do not touch the script until it closes the console and display the images, especially HaRGB Show takes a while, so don't be impatient as Pi can do strange things.
I put the script in a zip.
I would like to see results, and tell me if there is any fault, or if I can improve or what will happen.
Silvercup.
Edited again: Image Credits Jim Misti.
-
Looks really great silvercup! I like the results you're getting and also the GUI for the script looks really slick. One of the only scripts and processes I know with a preview -inside- the GUI. Nice.
-
Hi
This looks very interesting indeed :laugh: a couple of questions for you
1) do we need to prepare the Ha image in any way
2) Are you adding the Ha to the red channel
Regards Harry
-
Thanks Sander.
I only saw one problem adding Ha to the red channel, R channel "eats" G and B channels, althought you can recover with G and B saturation curves. So we must investigate a way to dim Red channel where you has G and B signal.
Harry, you only has to flatten Ha and RGB images with dbe if needed. The script do a linear fit between Red and Ha, then do the Vicent Peris pixelmath's with a little diference, in a third step the script do a max between original Ha and recombined HaR. Yes, I add Ha to Red.
Silvercup.
-
Hi
Thanks for the info All I need is some DATA to try this out
MAny thanks for your hard work
Regards Harry
-
Hi Silvercup,
Great idea for a script. Thanks for sharing it. Will give it a try.
Steve
-
Silvercup great job ! I also wanted to program something like this but time is always limited :sad:
BTW, feature scripts -> Add does not add it , I guess it misses the Feature Id etc declarations ?
I can open it and run it with F9 though. I can't wait to go home and try it on older images.
-
also trying to load it using feature scripts.....it won't recognize it. Any ideas?
Thanks
Carlos
-
Open the Script Editor, File Open, insert the script into it, open one of your images, press F9 to run the script.
BTW I tested it on a few of my previous images and did a prefect job. I had to spend a lot of time with try&error in order to reach this result. Great job Silvercup, thank you! (and please put the needed declarations at the top so that we can include it in the list of the scripts ;) )
-
Thanks for the Script, but the results aren't that great, all stars are red, there is no way to color calibrate the star colors and show good Halpha detail at the same time. Is there a way to do that(manually?) or maybe in a future update of the script?
regards
Daniel
-
Thanks John. I'm glad that script does his job.
Daniel if you get red stars there is something that you are doing wrong. I have done sereral combinations with diferent images and I never have obtained red stars.
Silvercup.
-
so what has gone wrong? I used the script exactly as you wrote...maybe my computer messes it up. Another problem is, I just can generate one HaRGB and say "ok" then it will open. If i do it once, and again, afterwards computer says something about memory access violation and PI will not respond anymore. So if i open PI again, i can use it once again, then have to save, reopen pixinsight, and use the script. Strange :surprised:
Will investigate further in this when i have the new 64bit operating system and the 12GB of ram...my computer at the time is a little limited (only 3,2GB ram @32bit).
regards
Daniel
-
Hi
Make sure all images on the screen do not have the STF active as this seems to play havoc with the script >:D
Also perhapes you might want to watch this vid as its what lies under the bonnet
http://www.harrysastroshed.com/Hagalaxy.html (http://www.harrysastroshed.com/Hagalaxy.html)
Harry
-
Hi all,
I enjoyed reading the posts on adding Ha to a nebula image. I recently acquired images of the Jellyfish nebula through RGB and Ha. I tried out Vicents technique using the process icons added through Harry's site (thanks Harry and Vicent for the pixelmath formula). I added a bit of a twist to the process by adding an additional 20% of the Ha signal to the blue channel and 10% to the green channel to account for Hb emission. All seemed to go perfectly with no change in the star colors. This technique is really awesome. This image is about 20hrs of exposure time with a 80mm f6 refractor from my suburban backyard.
Jon Talbot
-
Hi Jon
Always enjoy your images ;D
I too like this method , seems to work very well , just don't tell everone ;)
Harry
-
You don't have to tell everyone - just me.
Seriously, jon, I am still struggling with adding Ha and would be interested in learning a little more about how you processed this image. Did you use the script?
I'm presently reworking some data for the Flame Nebula that I acquired about a year ago. Last time around it was Photoshop all the way, and I didn't particularly enjoy the experience. I never really understood what I was doing (the usual condition for me). Now that I have divorced Ps in favor of PI, it will be PI all the way. I hope to report back with an image soon. So far I have been getting great results with MMT and HDRMMT on the luminance.
Dave
-
Hi Dave. Sorry to hear you are having trouble with the process. I just used the process icons that Harry had on his page with the pixel math formulas that Vicent supplied. You have to know your bandwidth of your Ha and Red filter. I used the Astrodon 5nm Ha filter and also use the Astrodon LRGB filters. The red filter is close to 60nm wide so thats the number I used instead of the 100 in the formula. It produced a nice continuum subtracted image with just a few bits of the brighter stars left. This is your continuum subtacted Ha that you combine with the red frame using the other formula. An important point is that you have to make sure you only apply the process on linear images before any stretching. Harry's video is also a good thing to watch if you havnt already done that. Once I combined the Ha to the Red, Green and Blue frames I just combined to color using the channel combine process. A screen stretch showed that I had a great Ha combined image without changing the colors of the stars.
I then processed the image as I would any other. Hope this helps.
Jon Talbot
-
Hi Jon,
I to have been using Vicent's PM formulas to make a continuum subtracted Ha and add it to the red channel with good success. Want to also add a small % of the Ha to blue as you did and wonder how you are doing this. Are you doing it with the 2nd formula and using the multiplication factor to determine the % added so if you want to add 10% then the multiplier would be 0.1?
BTW, great image of the Jellyfish.
Steve
-
Hi Jon and Steve,
I'm not sure I can relate Vicent's PixelMath formulas with the process icon supplied by Harry with his tutorial, but I'll be making another pass at this processing soon for the Flame Nebula, and I'll see how it goes. The continuum concept doesn't seem to be sticking to my brain.
Thanks for your helpful reply, Jon. I really appreciate this forum and the help it provides.
Dave
-
Thanks Steve.
The way I add a percentage of the Ha continuum image is: once I have the basic continuum image I make a two duplicates. One for the blue channel and one for the green channel. If I want to add 20% of the Ha image to the blue channel, multiply the linear Ha continuum image you just made for that channel by .2 using pixel math. Then take the resulting image and use Harry's process (substituting your Ha image in where appropriate) icon to add this image to the blue channel.
Hope this helps
Jon Talbot
-
You need to be sure to have your images open before trying to run the script. I have been playing with both methods and have to say the script gives you a good preview of the output new image. Thanks for the work on this and I'll be using it many times I'm sure.
Steve
-
Thanks for the advice, I'm going to try to use this for my M20 image (I'm entirely new to this and still trying stuff out, I've got no idea what I'm doing, so don't be surprised if I'm back soon asking dumb questions).
-
Actually this past week I've just tried using both processes as describe and had good results with each. I have to admit that using the script was a little easier but then Harry's video tutorial shows you what is actually being done. Make sure you try both ways and you'll learn a bit as well.
-
I tried Vicent Peris method of narrow band and broadband combination (http://www.pixinsight.com/tutorials/narrowband/index.html). When I created continuum map ... C = B / N , I obtained inverted image after division of regular images. When I divided LinearFitt-ed images, I got totally black image. I don't know what is wrong. Please, any suggestion?