Please can I validate my RGB + Ha workflow?

[mads0100]

Check out Straton to remove the stars.  It's hard to find; german software.. there is a link on Astrobin to it.  It costs 20 bucks I think.... It removes stars (very well I might add) from the picture and will make them look signficantly better.

[topboxman]

Is this what you are referring?

http://www.zipproth.com/Straton/

Peter

[jerryyyyy]

Got it, but is the point to fix the gradients on a nebula with the stars removed, then put them back? 

BTW, still not clear if you should ever run a linear fit on any of these componnets RGB components and/or Ha?  I have found that I can set the media level in the MaskedStretch and life works out better if I use the same median for the RGB image as for the Ha. 

SOrry for the dumb questions... 

[topboxman]

To be on the same page and in case I mis-understood you, don't put the stars back from Ha image. Remove stars from Ha image and run the HaRGB or NbRGB script with combined RGB. This way RGB stars will be preserved and not be ruined by stars from Ha image. I have been planning on trying this but I have been too busy. I will try removing stars from Ha image with PixInsight and see what happens this weekend if nothing else interferes with my plans.

As for processing individual channels like DBE or Linear Fit, I am not sure and I think it's entirely a personal preference. IMHO, running DBE on individual channels or combined RGB may not make much difference but try it for yourself and let us know. Personally I use DBE on combined RGB and Luminance separately but I have not yet tried DBE on individual R, G, and B channels.

Also, after combining Ha with RGB using HaRGB script, next I run DBE. I haven't yet tried DBE on Ha alone. If you run DBE on Ha first, then you might want to run DBE on combined or individual RGB before HaRGB combination. Just make sure the timing of running DBE is done correctly.

For Linear Fit, I have seen Linear Fit demostration using non-linear (stretched) images but not on linear images. The word "Linear" in the title "Linear Fit" may be mis-leading. Once again, try it for yourself using both linear and non-linear images. The results may be enough to tell you.

Peter

[jerryyyyy]

Thanks, that explains what to do... you add the de-starred H-a to the RGB and avoid pink stars!  It is 15Euro and I tried it on one image and it worked, so will pay up and see if it helps. 

i have been trying a few different combo of DBE as it was slow at the office today.  Doing the LinearFit will definately change everything, so probably not too advised.  Frankly, I seem to get better results putting the image together and then doing the DBE... this is two less overall manipulations. 

I also seem to get better results with the ColorCombination than LGBCombination....

[topboxman]

I also bought Straton star removal tool. It seems like in order for this tool to work well, it works best if the image is in 16 bit unsigned FIT format rather than 32 bit floating point FIT. Straton tool does not seem to process image in 32 bit floating point FIT format very well. It takes a long time (almost 15 minutes) with 32 bit floating point and does not appear to work at all. If your original file was saved by PixInsight in 32 bit floating point FIT format, it would be best to save this file as 16 bit unsigned integer before removing stars with Straton tool.

But I am having a heck of a time combining de-starred Ha image with RGB using NbRGB script and process from then on. I am still working on two different DSOs containing both Ha and RGB and hoping to figure out the best way to get HaRGB image using RGB stars.

If you figured it out, please post your procedures.

Thanks,
Peter

[topboxman]

No matter how much stars I removed from Ha image, I still cannot get the stars to look more natural from RGB image when combining de-starred Ha image with combined RGB image using NbRGB script. The Straton star removal tool seem to affect the background data and also makes it harder to do proper Background Neutralization and Color Calibration.

I am not having good lucks using PI Star Mask to reduce/remove stars.

Any advice?

Peter

[pfile]

how have you created your mask for star reduction?

[topboxman]

First, I got the idea from Harry's Astro Shed for Star Mask for Linear Image:

http://www.harrysastroshed.com/linmask.html

Then I had to tweak a few settings of PI Star Mask including Midtones and Noise Threshold. I believe the biggest trick I learned from Harry is to use Histogram Transformation and move the Shadow slider to the right until all background noise is gone and the background is pitch black while the stars are still bright white for Star Mask image. The blacker the background for star mask the less it ruins the target image and makes color calibration easier later. It was not easy for me and I spent half a day figuring it out. Personally I would not recommend Straton Star Removal tool unless I am doing something wrong. This tool seems to mess up the background and content of DSO that made it harder to post process DSO.

Also I believe it's better to use multiplication factor as low as possible for NbRGB script to avoid messing up the stars and background. So far I've been using 0.9 multiplication factor.

Probably the biggest difficulty of all is the signal strength of Ha image is much weaker than RGB image and that made it even more difficult creating a good Star Mask for Ha image.

Because I am deaf, I can't hear Harry's narration so I had to make the best educated guess of what he is trying to say.

Peter

[pfile]

if you have a field with a lot of small stars (like the veil nebula or such) you can extract L*, then using AtrousWavelets, remove everything but wavelet scales 1 and 2. then stretch the image with HT so that the stars are visible. it might be helpful to use a preview and look at the preview in the RT viewer while stretching, since the stars are so small the full image just looks black in the RT viewer.

then, you can reset the AT tool and remove just scale 1 to smooth the image a little bit. you may need to adjust the stretch at that point.

you can also experiment with how many scales to preserve or delete in both steps to try to come up with an acceptable mask. you can also hit the mask with the MorphologicalTransformation set to Dilation (just a small amount) to open up the star mask a bit, if necessary.

sometimes it's hard to get a good star mask using StarMask when the field is just super crowded with tiny stars.

not sure if that applies to your situation but i thought i'd mention it.

rob

[topboxman]

Thanks for mentioning. Interesting method using AtrousWavelets for stars removal/reduction.

Peter

[jerryyyyy]

I am in the same boat (wasting time on a Sunday afternoon).  I paid for the software and it did a great job removing the stars, but when I used the new variant of Harry's H-a script, I have too much h-alpha signal... I am down to 0.05!   Maybe there is just a load of H-alpha in M33. 

[topboxman]

I think Straton Star Removal tool ripped out too much data from DSO signal and background making it extremely difficult to process from then on. I think you are better off to do a manual process of star removal using a few PI tools including Star Mask. It is a time consuming process. In order to make your life a little easier, concentrate on large stars when making Star Mask. Smaller stars seem to be irrelevant when making Star Mask. That's what I did when trying to process Bubble Nebula. There's so much nebulosity surrounding Bubble Nebula, this was a very difficult task of removing stars especially if the image is still linear long before stretching with Histogram Transformation.

Bubble Nebula before Star Removal using PI mask:

https://www.dropbox.com/s/29742yw6a1xi3cf/Bubble_integration_Ha_Crop.fit

Bubble Nebula after Star Removal using PI mask:

https://www.dropbox.com/s/4osgre470dycari/Bubble_integration_Ha_Crop_ReducedStars.fit

You will notice the large stars were removed/reduced and smaller stars were dimmed but I think it's enough for further processing when blending Ha with RGB.

If you can do a better job removing/reducing stars than I did without sacrificing the nebula part, please upload your result of my image.

Thanks,
Peter

[jerryyyyy]

Looks to me like your process also took a lot of detail out of the bubble itself.  Do you have RGB files to which you are adding the H-alpha?  I see I did this before I had a CCD camera.  Looks like it is in imaging range for me.  Think I will put it on the list. 

I looked at images on Astrobn and found this one:



Detail are missing but looks to be a Hubble Pallet.  I asked the person what filters he used.  May be very hard to add the H-alpha in here... but most of the other images on Astrbin seem to go that way.  But does not look easy. 

[topboxman]

This is a calibrated and stack of RGB image of Bubble Nebula I am trying to blend with reduced stars of Ha:

https://www.dropbox.com/s/j827atif1sxvbeo/Bubble_RGBComb_Crop.fit

Calibration, Star Alignment and Image Integration were all done with PixInsight. You can thank my heavy light pollution for showing ugly gradients in RGB image.

This is my Bubble Nebula processing results including HST, Bi-color, HaRGB and RGB:

http://peternagy.smugmug.com/Telescopes/Nebulae

Look at the texts for details of each image.

I just compared my Ha Bubble before and after star reduction with PixInsight and there appeared to be little differences as compared to what Straton Star Removal tool did. I used STF to do the comparison and with equal stretch, there were very little differences using my Star Mask.

Peter