Anyone out there know how to stretch, curve, or otherwise manipulate an image so as to enhance the nebula without causing star bloat, and/or reduce stars without distorting the nebula?

I have a piece of the Veil Nebula that I have been trying to process, but if I get the nebula pretty good, the stars are bloated to the point of detracting from the nebula.

It seems like my first processing effort is the best. I've tried to reproduce the above image, but I always seem to overprocess. When I go back and try to process it again, I end up over processing and lose the quality. I try to squeeze too much out of the image. Does anyone else experience this?
 

...just a newbie lookin'  for some advice from more experienced astrophotographers.

Use a program like GIMP or Photoshop and, under the color tools, select curves.  Adjust the gain in an S shape, pulled down on the left and pulled up on the right.

CFB

Lothar:

The best way to do this is by using curves. 

Start with a raw image, that has yet to be stretched...meaning, it's mostly black with a few stars appearing as dots.

Then, open curves and bend up on the lower left side of the line (those values represent the faint stuff).  When you bend the curve, put an anchor about half way and straighten the line up the the upper right corner.  Don't allow the upper values to bend...this will allow the shadows to come out without bloating the stars.

Repeat this a couple of iterations, until the background starts getting bright, then open levels and drag the left slider to the beginning of the histogram "hill"...actually, just before it, so you don't "clip" it.  

Then repeat the curves as necessary, resetting levels as needed, until you are happy with it. 

Thanks for the tip!

I still struggle with all my images, trying to get the most out of them.

Is there a process you follow after getting to a flat image?

Allow me to explain...

I have a DSI color, and set the camera software to save 4 files for every exposure (LRGB)

I combine each of the separate color exposures in MaximDL, then open them in Photoshop with the FITS liberator, pasting them into their own layers

Each layer is then converted so all layers are colorized and transparent.

After some Levels adjustments, I flatten the image and begin stretching with the Curves and Levels controls. This is where I have problems. I can never seem to get the background dark enough to be pleasing. The tip you gave certainly does help, but I'm thinking I can do more before the image is flattened.

When combining images, do you adjust the levels before stacking them?

When opening in imaging software for processing (eg. Liberator and/or Photoshop), do you adjust the levels before converting each layer? Afterwords? 

I guess my basic question is, do you adjust levels in each step in the processing, or wait until the image is flat to really stretch it?

There are so many ways to alter the image, it's hard to know the best way, or is that just experience with processing images? I know it's as much Art as Science, but I think I'm talking about the Science part.
 

1. Calibration/Reduction

2. Registration

3. Combine

4. Post-process

Levels and curves are in stage 4...don't touch the images until then.

Once you have combined the images to form your RGB, you should probably work each channel individually, unless you have normalized the backgrounds (made backgrounds for each channel the same).  

The goal to getting good color balance is to have identical histograms for each of your RGB channels.  For example, here are M13 and NGC6888 screen shots showing what I'm talking about...

http://www.allaboutastro.com/articleimages/M13histograms.jpg

http://www.allaboutastro.com/articleimages/NGC6888histograms.jpg

Always have your histograms open when you do any processing step.  Always protect the hills.  Do not cut into the hill on the left (that darkens your background too much and loses data).  Do not cut into the histogram line extending to the right (that's your stars).  Cut too much into that and you will bloat your stars.  In fact, I like to open levels and slide the right output slider to the right a bit...it gives you a gap to the right side of the histograms so you can see if you cut into that line. 

You'll use levels to keep the hill in the same spot, using the left input slider to snug into the hill without cutting into the line that extends out to the left of it.   In curves, as I said in the earlier email, you'll bend the curve on the lower left and flatten it out towards the top.  This will bend the lower (shadow) values in your image.   The effect of this on the histogram will be a gradual reduction in the slope of the left side of the hill.  It will flatten out a little.  If you do it right, you'll get that flattening without seeing any movement on the right side of the histogram, particular where the star line is. 

After each iteration of curves, be sure to reset the blackpoint in levels.  

You can do this on the RGB image if the histograms (color balance) is good to begin with.  Otherwise, you'll be better off working the channels separately to try to make the histograms equal.

Hope this helps,

One more thing.  You mentioned that you wanted have a dark background that is pleasing.

Be careful there.  

My images really started to improve when I realized that their is something worthy to see in every part of the image.   Space isn't black.  

So, when you process, try to keep the background somewhat brighter than you think, and look for goodies...dust, faint galaxies, etc.  

The problems is, when you just start, is that your images usually aren't very long...so you tend to need to darken the background to hide the noise.   So...a tip...learn to image long enough to where you can have the background bright enough to see those goodies...or at least be able to utilize forms of noise reduction.    If you are shooting less than 10 minutes total, or so, you'll have a difficult time making those backgrounds smooth without just making them black.  

Here are some screen shots of the curves and levels I'm talking about.

First, here is the type of curve that will bring up shadows yet preserve your stars...

http://www.allaboutastro.com/articleimages/rosette-log-curve.jpg

Yes, it brightens the background, but that will be reset by the following levels adjustment...

http://www.allaboutastro.com/articleimages/rosette-levels-reset.jpg

Notice the red arrow.   I just dragged the slider a little closer to the hill.  Go too close to the hill and you'll overly darken the background. 

At the end of the process, you might end up with something like this...

http://www.allaboutastro.com/articleimages/rosette-final-adjustments.jpg

Notice two things.  One, there's a lot of information in the background that would have disappeared if you left your background black.   Two, notice the histogram, how it now gently slopes on the left, yet the hill doesn't extend much further on the right than it did.  Because I adjusted curves, the histogram is redistributed, meaning that the shadows now have more of the values than before.  Essentially, I've just awoken half of the image that wasn't there in the first place.

In that last one, I did some other tricks to increase contrast in the details, but that's something you'll have to learn later.  As a result, the final image is something that I could live with...I'd likely darken the darkest parts a bit more, but that's being picky.

Remember, you aren't shooting empty space.  You are shooting a landscape in space.   There's a big difference.

 Wow, thanks for all the info!

I never even thought of comparing the histograms between color layers.

Sometimes I would get 3 separate 'hills' when stretching the flattened image. Now I know why.

I will try reprocessing the bit of veil nebula I imaged recently.

Thanks again! 

Fascinating thread! I will definitely be checking this out once I start getting into long exposure shots, maybe this fall. Meanwhile thanks for the info so I can start reading up on it.

Good to see you, Steve....been too long.  

Hopefully, I'll spend more time out of my hole! 

And, BTW, I can't imagine you'd have much more to learn.   I've never seen anybody get the data you get from so little exposure time!

hi jay- hows it going?- you know, this thread gave me an idea, i might add some nebula enchancing ATN's on the the next issue of spitfire's scopestuff- LOL

I followed your advice and re-processed the bit 'o veil I imaged.

Here is my first attempt, before reading through your tips...

Not bad, but here's after following (and practicing) the techniques you shared...

Although it may not be that apparent, I think it's a definite improvement.

Thanks Jay! 

LOL, I'd expect that from you, john.

Good to see you!

Oh, a definite improvement!   You've managed to show the colors a bit more true.  Your first effort did do a better job of controllling the stars, though...not bad at all!

If you'd like, send me your raw data to my email at jballauer at gmail dot com.   I'll work it up for you so we can see the potential that's there. 

 Thanks. Thanks again for sharing your knowledge!

I stretched the 2nd attempt much more than the first, so I think I kept the star bloat to a minimum.

Do yo want all the exposures? 109 FITS exposures for each; L,R,G, and B. That would be a lot.

How about the 4 FITS files that are the product of combining each of the color exposures?

Exactly.  Just the master files for each of the channels.  

 Images sent.

Okay, here's what I came up with...

In short, you've done a pretty good job of processing it yourself!

Of course, the real weak point is not your processing, but rather the data.  You'd be amazed at how easy it is to process good data.   In fact, I'd encourage you to try your hand at processing some of the data that Jim Misti has made available here...

http://www.mistisoftware.com/Astronomy/index_fits.htm

Now, back to the data...

I'm wondering about the parts of the image I had to crop.  The top part of the FITS files you sent me were striated and unusable.  I wonder what would have caused that?   As far as the vignetting, I would make sure there is no direct lighting around, including the laptop.   For 30 second images, your background levels are extremely high, meaning you most certainly are shooting in some bright conditions...assuming the master files you sent were averaged rather than added. 

Yes, flat fields would certainly help.  Try shooting sky flats in the early morning...or doing "T-shirt" flats, stretching a t-shirt over the aperture end.   However, I would resolve to figure out why you have that much vignetting in the first place.   I'm not sure it's mechanical vignetting (something cutting into the light cone), but rather than you have a hot spot in the optics (all scopes have such a spot) and it just becomes more pronounced with light pollution.   Though, again, that striated part concerns me...don't know what could have caused that. 

Just a few tips once you acquire the data...

- Don't use luminance data if it doesn't help the image.  Honestly, the green channel was your best data...if you are going to use a luminance, just duplicate your green channel.   I know your L layer was probably synthetic, due to the single shot color camera, but I'm talking conceptually here.   When you process with an L layer, you are telling the software that you want your details FROM that layer.   So, you'd be certainly that your L layer has details worth detailing!

- Be picky about the images you use in your stacks.   If you shoot 90 images per channel, chances are VERY likely that MANY of those images are blurred, with oval stars.   In that case, including them actually kills your S/N level in the image.   You are better off selecting the best 10 or 20 of them.  

- I don't know what type of reducer you are using, but be sure you have the right distance between the reducer and the chip...if not, then you won't have the right reduction and you add aberrations to the image.   More than likely, you have one of the .5 reducers that screws into the eyepiece, which kinda negates the point.   However, it's good to double check that.

- Don't fall in love with the reducer.   When shooting single objects like this, it doesn't "speed" up anything.   You are often better off shooting natively.   Now, if you are doing it because you need the forgiveness offered by that image scale, then it's a different story.  But the point is, if you have the gear to shoot at the longer focal length, the results will be better on an object like this if you do it.


It's really good to work on your processing...you are beginning to understand the importance of the white and blackpoints, and what we call "logarithmic curves" - namely, boosting the shadow areas more than the highlights.   But I think you are at the point where you need to work hard to get better data...that's what will really improve your images.

Hope this helps!

 My mount is Alt-Z, and my scope does all it can do just to keep the object in view, meaning it moves around quite a bit in an imaging session. When stacking, the registration image has the object at the middle of the view, but the object in the images towards the end of the session are at tht bottom of the view. That's why the bottom of the images don't have much data.

I excluded a good many exposures due to field rotation.

I have to tweak the anti-backlash settings to make the movement steadier. I've done it before, and it does help. I was experimenting with some telescope control software and inadvertently reset the anti-backlash settings to 0.

The vignetting isn't always an issue. Some shots I've taken have very little or no vignetting, with or without the reducer, and others it is very bad. It's probably like you said, light pollution or stray light from my laptop. I will try the T-shirt flat to see if that helps.

thank you very much for taking the time to process the images and provide feedback!