Example of creating a full LRGB Image

S

sreilly

Well-known member
I may have not seen this information, but for a beginner to assemble an image from the begining is confusing. My usual process using MaxIm or any other image program (CCDStack, CCDSoft & MaxIm) I calibrate raw images (ST10XME) using Darks, Flats and Bias frames. Using CCDInspector, I'll cull the usable data using data with FWHM of 2.5 arc seconds or less. All images are aligned using Registar. At that point master red, green, blue and luminance images are created, most of combining im MaxIm using Sigma Combined by means of Russell Crowman's plug-in. Now, the RGB image is made in MaxIm and saved. The master luminance and RGB images are saved as 16 bit and opened in Photoshop where curves and levels were used to bring out the faint details. Selective sharpening on the luminace data and then combining the color and luminace data. Noise Ninja may be used if noise is present for the background.

This is my basic image process and I'm looking for some guidance in using PI for these processes.

Thanks,
 
Steve
I just went through this learning curve... :'(
Harry was a great help,and even did a little video on it...
It in the forum,under "dumb question about LRGB..." and "another dumb...".....(!!)
(I was referring to myself,I assure you !!)

He has a video of it in his collection...
One produces a separate set of images...the RGB,and the Luminance...
Next step is in color spaces..."LRGB combination"...
Quite fun actually !!

Dave
 
Hi Steve,

You should also keep in mind that the forthcoming 'next' release of PI will give you greater control of all the calibration steps, so keep that in mins as well. I already now perform ALL of my 'calibration', 'registration' and 'alignment' in PI - and have never experienced such excellent results in other software packages (admittedly, I have NOT invested in the likes of Maxim or CCDStack, and so cannot give you a direct comparison with those packages - however, my feeling now is that if PixInsight can achieve such excellent results, why would I need to trouble myself with those packages in the first place ::))

Cheers,
 
For now i acquire my images in Maxim and calibrate with ccdstack.  WIth the images in 32 bit floating point I use the Star alignment tool to align all my images to a selected registration master at one time.  Then I use Image integration (settings I use are Average, Adaptive and Scaling & Noise to combine the different filters to make my masters: Pixel rejection is windsorized with scale and zero offset.)  Then combine your masters in LRGB combine.  I then use Dynamic background extration while the image is still linear.  Now open STF and use the sliders to do a nondestructive stretch to veal your image.  From here you use Histogram, Curves, etc much like other programs.  SO this is a basic start.
 
Thanks Jack, that was more helpfull. Niall, MaxIm does the image capturing part extremely well and is also required for ACP to work. As I automate as much of the data collection as possible, including auto sky flats, darks and bias frames, MaxIm is the software of choice. Only if and when PI takes on data acquisition, will it be able to be a complete imaging program. My impression is that PI is interested in image processing, not data acquisition so other software is needed.

The thread on CCD-Newastro has been somewhat passionate but not much help. The lack of a help file or manual is a real handicap to new users of PI such as myself. That said, theer are many features well documented in MaxIm that aren't easily understood. While deconvolution is a strong tool, the defination is there and some very basic instruction but not enough to really help. The same could be said with ddp in MaxIm. It wasn't until a few years ago when Adam Block advised me of his filter setting and method of getting the mdi-point and background values did I ever have any success. That said, the manual has been extremely helpful in many other functions.

Thanks Again,
 
What you probably need is the Screen Tranfer Function (STF) process. Allows to push the image visually without changing it.
Georg
 
Another issue is the allowed weights for RGB. My ratios are .67:1:1.39. Most all packages I've seen allow this type of color ratio fpr filters but in PI, 1 is the highest value allowed. I've always seen the green channel being referenced as 1 and the red and blue accordingly. This seems to be entirely different in PI. Now it seems that my ratios need to be .48:.71:1 which is not the normal method, to my knowledge. Is there a reason that the values must be limited to no more than 1?

And please explain the difference in usage of apply (F5) and apply globally (F6) in the LRGB Combination screen? The Target field under Channels/Source Images is also not explained.

Thanks,
 
I think if you just make your largest value 1 and adjust the others by dividing them by 1.39 you should be good to go?  I do not use G2V myself but this should work.  I always use apply globally and it creates a new frame entitled "Image 4" or something like that.  I do not use target field nor do i adjust the sliders for saturation etc.

Georg is correct on the STF.  Another "pearl" is to adjust the sliders to get an image, then click the RGB boxes on the upper left and individually set the black points and gamma.  If you get a good color mix then drag the small triangle in the lower left of the STF to the bottom task bar of the Histogram tool and it will transfer the changes to the Histogram which then allows you to drag the Histogram triange to the image to execute permanent changes
 
Hi Steve

I guess there are several ways of looking at the colour / balance issue which I see like

1) get the colour as close as you can with adjusting the weights as you say ( it might be different in PI to other packages , but we try not to hold it against them  ;D)

2) Just chuck all your colour images in at the default weight ( 1 ) and then carry out  colour calibration to sort out the correct balance , which is what I do

Also don't forget the STF only alters the appearance of your image on the screen , no data has been altered  :eek:

As for the apply And apply globally I see it like this

You can only use APPLY on a single image , so if you use it in the LRGB with separate LRGB channels it will not work ! but if you are using the tool to apply a lum to a RGB image it will work.

The global apply is applying the tool to multiple images so you would use this also say in the image registration tool  ;D

I would use the
target selector if say I had more than one RGB image open , but as I rarely do it does not get used very often and I leave it blank most of the time.

Some of the questions you are asking are covered in my newbie video's if you would like to watch them ,but I understand if you find my voice irritating  ;)

Regards Harry
 
Thanks Jack. This is going to take a lot of pratice as expected. I am following the videos from PI as well as Harry and others but what gets lost is the assumed base of knowledge. A prime example is the DBE PI video part 1 where the Histogram tool is being used. I can see that there are other actions taking place using the sliders and some type of zooming maybe? Even with a seperate monitor showing the video and a seperate monitor to try and follow the examples, too much is lost.

This reminds me so much of learning a computer language. Telling someone to take a shower and then really telling them how is not so easy. You can't assume anything. Open bathroom door, enter bathroom, close bathroom door, remove all clothing, open shower curtain, turn on hot water, turn on cold water, adjust to comfortable temperature, close shower curtain, lay down bath mat at shower, hang towel over shower rod, open shower curtain............. you get the idea. And this is actually easier unless there are assumed actions, such as the shower curtain opens differently with the right hand versus the left hand or entering with right foot first and so on.

Once you've done the proper process it's easier each time till it becomes almost automatic. It's the very first times that will frustrate you to no end. I've done image calibration for over 12 years with MaxIm and it's something I don't even think about. It's become a automated mind process that requires onlt the slightest attention. The same with aligning and creating the masters. I've done it thousands of times that minimal thought is needed, except when there is something that turns up as unusual and needs special attention. Now mind you, I'm not saying the whole process is like this, just the basic construction. But I can say I usually spend less that an hour on anything beyond the basic image, meaning curves and levels and other PS adjustments for a finished images, or what I consider finished.

What would be a big help is a basic start to finish lesson that covers image registration (no problem), image combine (no issue here), RGB/LRGB image compiling (no issue), using Histogram stretch, screen transfer function, DBE, curves and so on explaining when to use and why. The order of the process is important. Is there a function that simulates layer masking? Lets say that once I figure out how to use the Wavelets Transformation tool, I want to use only the part of the image that say involves the core of my galaxy, how can I reveal onlt that in a newly combined image?
 
You can extract the lum layer using the Extract Lum icon on the top task bar (5th icon from left) or use Channel Extraction tool and select CIE L a b and uncheck a and b.  Then use the histogram to adjust the mask.  Apply the mask by right click on workspace and select the mask (invert it if you like).  Or if its for stars use the StarMask tool.  GOod luck  Got to run!
 
Hi Steve

PI does not at present have layers , but I think mr Juan has said they are in his grand plan  ;D

Thanks for your comments about things as we can forget what its like at the beginning >:D

I did try and put my video's in a basic order that you would use and did not make them techy so people could follow them , please say if I missed the goal


Regards Harry
 
Quote Harry "Some of the questions you are asking are covered in my newbie video's if you would like to watch them ,but I understand if you find my voice irritating "

Not at all Harry. I've downloaded all the listed videos but haven't had a chance to view them. Probably best to have some logical order to watch so I don't put the cart before the horse. A posting on CCD-Newastro says it well "Hi - I'll get it started - being a new user I would like to see
something as simple as a "Flow Chart" showing all the menu processing
options and where in the image processing it was recommended to use
them. If something like this was available in PDF format then I could
print it off and stick it on the wall in front of me. That way I know
what procedures to use on linear images and which procedures to use on
stretched images. Once a procedure has been allocated a specific
position on the tree then informed discussion will no doubt follow
because not everyone processes images the same, all have different
ideas, but at least there is a starting point.

Cheers
Geof Wingham"

That's an excellent idea and really a great way to start.

Thanks,

 
You must log in or register to reply here.
Back
Top