Star Removal

[mbushee]

Hi,

So I have a question regarding star Removal. I know that this is done in the heal module after de noise typically. My question is it possible to remove the stars before processing is done? I currently have a Altair quadband filter for OSC camera. I typically image from a bortle 6 and I got the filter to get more narrow band images. My issue has been in most images I end up with a huge starfield which I understand that's a product of the filter. My latest image is of Sh2-132 (Lion Nebula). The stars have been making it difficult for me to pull out the Narrow band details I would like and when I use the shrink and super structure the smaller stars almost end up looking like noise. Should I be playing with a different setting maybe?

Anyway I really don't like the star colors from the filter so I'm planning on layering in the RGB stars, but I'm having a tough time getting rid of the stars at the end during the heal module so I don't know how I would get rid of them in the layer module. This is why I'm asking if the be stars can be removed in the beginning before I start running through the modules.

It doesn't help that this target is right in the milkyway so more stars then some my other NB images.

I've attached something I ran through quick it's definitely not my final image but I just wanted to show what I'm talking about in regards to the stars. I basically want all NB stars eliminated so I can put RGB stars in. I also don't want starfield to interfere with NB details.

The image below is 15 hours of data with 300s subs with all calibration frames (darks, flats and dark flats). I have started processing this again and it does look better but I'm still having trouble removing the stars and the stars make the final image look off. I'm going to be getting some RGB for the natural stars tonight. So stars in the image are from the filter.

https://photos.app.goo.gl/JJ9spmxeSDfxUuov9

-Matt

[almcl]

Hi Matt

I had a play with your image, which is a target I have tried to process unsucessfully a couple of times.

I don't know a way to control the stars in StarTools and still bring out the nebula, maybe others will chime in? But I did try feeding your jpeg to Starnet++ and got this:

Lion Nebula starless.jpg (172.66 KiB) Viewed 3954 times

If you do a very gentle stretch of your original image you could layer in the stars from that to get something like this:

Lion Nebula toned down.jpg (389.17 KiB) Viewed 3954 times

[mbushee]

Hey almcl,

My initial run through I think I over saturated it a bit and I also had it a bit softer than I would like. My latest run through is a bit sharper but, it still looks noisy from trying to push back the stars in the shrink and super structure module.

I was hoping since narrow band really has no need for the stars I could take them out prior to doing any actual processing. I have no idea if that's possible in Startools or not. I could try out star net or something, but no idea how that would affect the photo to edit in Startools. I usually do 100% of my processing in Startools.

I do like both of your results more than mine. Did you just do another stretch in Startools with autodev or did you use another software?

I can post my second processing tonight when I'm on my PC I can also provide the tiff if that would help I just didn't bother with this on because I personally think it's to soft and I bit oversaturated that's me though. Hopefully I can get 90 minutes of RGB tonight and eventually get the other stars out and the good ones in still doesn't help with trying to process the NB on it's own.

Thanks for the help though! If it would help I will also post my gear just in case I'm just using bad settings for this target.

*Edit

Here is my latest version both a tiff and a jpeg. I managed more detail, but still to many stars.

https://photos.app.goo.gl/JzcFgQH7f4LzgZc89

[almcl]

Did you just do another stretch in Startools with autodev or did you use another software?

I took your jpeg (as I didn't have the original fits file and just dialled back the gamma in StarTools' FilmDev. Not my preferred method, usually I just do a very gentle stretch in AutoDev and once happy with the result extract the stars use the layer module to combine them with the starless image.

Here is my latest version both a tiff and a jpeg. I managed more detail, but still to many stars.

I think you could use the DimSmall preset in SuperStr to reduce the stars some more in your latest one. I've tried a quick go here:

lion c dimsmall.jpg (322.02 KiB) Viewed 3922 times

[mbushee]

I guess I'm confused at when I should be extracting the stars. I feel like if I have to wait until the end it makes the ionized gas look all grainy. I feel like the gas is sharper and more detailed in my second photo, but the final product with no stars actually looks worse then my first one haha.

I've been struggling on this one with my second Autodev after the wipe. I usually click redo global stretch (button on the left) the second image I didn't even set a roi I just left the stretch from the wipe to try and get as much dust as possible stars make it a little difficult though. Should I be adding an roi or just useing the stretch from the wipe. Just not sure which way to get best results.

Anyway when in the process do you typically try and remove the stars? After the denoise? I feel like it leaves in tons of cyan where some of the brighter stars are shouldn't that be removed also when inserting new stars or is that in the layer module?

Side note I'll throw my fit file on Google drive if you want to have a look. I spent 15 hours on this I just want a image I'm happy with lol.

Thanks

[mbushee]

Here is the Google drive link.

https://drive.google.com/drive/folders/ ... F-HfYH76oz

Thanks for any help you can provide. I've only been doing this since March, but I've gotten better, but I'm just struggling on this one for some reason and it really bugs me lol.

Still hopeful we can somehow remove the stars in the beginning as I think it would help to just edit with no stars.

-Matt

[almcl]

I haven't found a way of extracting stars at the beginning and have to resort to full processing. Then either use the heal module to remove the stars, possibly followed by the filter module to subdue the more prominent ringing artefacts, or to feed the processed image to other software to remove the stars followed by putting back rather more subdued ones.

If anyone has a solution it would be great to hear about it?

[admin]

If you really need/want to process a starless linear image (calling it a dataset at that point may be arguable), have a look here;

viewtopic.php?f=7&t=1476

Please be mindful of the fact that the more stars are in your image, the more such software "makes up" out of thin air, so keep this in mind when you decide (or not) to rely on the resulting nebulosity.

In general, processing busy star fields (without doctoring the image) win widefields, in the following order, benefits from;

• Using Ha only (if available) for luminance
• AutoDev with fine Ignore Fine Detail < greatly increased (this progressively ignores fainter stars when allocating dynamic range to the shadows, and only allocates dynamic range to the shadows based on larger scale structures)
• Deconvolution to reduce stellar profiles
• Shrink (with deringing increased) to reduce stellar profiles further. Also consider using the "AltMask" preset in the AutoMask generator for more comprehensive star detection. Also consider using "Unglow".
• Super Structure (reduce Airy Disk Radius for widefields)

Stars are important, as they do give important visual cues as to emission and dust concentrations. Application of StarNet++ and similar tends to destroy this correlation that the Super Structure module seizes on.

In this particular dataset, I did notice, however, some misalignment of the channels, which may be making things harder to process...