Controlling stars

[almcl]

Starting this thread in response to a comment by Jeff in the Special Techniques thread.

Jeff wondered if there is a simpler workflow to keep the stars under control while teasing out the nebulosity, that perhaps didn't involve Starnet++?

I have had a go at the posted jpeg in that thread (1st image here) and put my effort below. Not sure how much of an improvement it is?

If it's of any use to anyone I'll explain the steps which basically involve creating two differently stretched files, one with just stars and one with just the nebula and combining them in the layer module.

All the processing was in StarTools. It was quite a 'quick 'n dirty' process and a much better result could be achieved starting from the original stacked but otherwise unaltered image and taking a bit more care over masks &c.

Original:

The original jpeg

Starless+starmask_Bode_sGalaxy.jpg (525.76 KiB) Viewed 9127 times

Reprocess:

The reprocess

Bode_sGalaxy reprocess2.jpg (324.39 KiB) Viewed 9127 times

[jlh]

Hi,
Thanks for moving my question about separate processing techniques for stars here. If it would help here is a link to the linear FITS file for this image: https://drive.google.com/file/d/1jtelib ... sp=sharing
(As mentioned before, this is not great data).
Jeff

[almcl]

Thanks for sharing your original data, Jeff.

I have had a very quick play, using the technique I outlined in our PM exchange.

I seem to have lost a lot of detail and the colour has come out differently, perhaps as a result of a rather too quick use of Wipe, but see what you think?

A blend of two processes

Bode_sGalaxy blended.jpg (253.73 KiB) Viewed 9087 times

[LuckyEddie]

If it's of any use to anyone I'll explain the steps...

Yes please!

[almcl]

OK, here we go.

It centres on the technique originally described here:

viewtopic.php?f=7&t=244

by Ivo for separating the stars from DSOs. This is also referenced in Guy's list of Special techniques

Starting with the original stacked image, Optidev, Crop and then wipe.

After this apply a very gentle stretch in Optidev, perhaps putting the RoI on a star. Follow the usual processing steps, as a minimum perhaps [Contrast > SVdecon > Color > Shrink > Superstr] to produce an image of nicely coloured small stars and then turn off tracking and apply the 'Using the Heal module to process stars independently' special technique to generate a stars only image. Save the 'stars only' image.

Now reload the original stacked file, (StarTools will remember the crop setting from before, so the images will align properly) and this time process it to bring out the nebula/DSO, not worrying too much about star bloat. When happy, turn off tracking and again use the special technique but this time save the starless image.

Work on this starless image to remove any remaining artefacts - the isolate preset in Superstructure in conjunction with a good mask and 40 pixels worth of mask fuzz can be helpful although less than 100% strength is often enough - and the reject option in Filter with circular masks round any errant glows can be used against star halos. Once the DSO is sitting on a satisfactory background, launch the layer module and open the previously saved 'stars only' image. Use a layer mode of Brighten or Add to combine the two layers. Blend amount can be adjusted to taste if necessary.

This works better on large well defined objects such as galaxies but can result in loss of detail such as smaller background galaxies. PGC 28757/Holmberg IX has practically disappeared in the rendition of Jeff's image above, for example. More careful masking during the Superstructure/Isolate part might have saved it. Getting a suitable mask on supernova remnants or reflection nebulae can also be tricky.

[LuckyEddie]

Thanks almcl. I'll have a play with that on some of my old data.

[dx_ron]

Having nothing better to do (that I wanted to do, that is), I played around with this with my recent NGC 891 data.

Here's the earlier straight-forward version:

NGC_891_390x120s_1600_v3.jpg (480.47 KiB) Viewed 423 times

Here is a first go at Heal-based separate processing:

composite_v2.jpg (297.5 KiB) Viewed 423 times

It's pretty terrible, but more in the sense that I went way overboard with the "small stars" idea in the star-stretch version. So it looks like I took some scissors and cut out a few galaxies and plopped them done on a photo of a star field. Why not re-do it more sensibly before posting? Because it took way more time than I want to repeat now (though I saved most of the masks, so hopefully it could be repeated more easily).

Will I ever use this for real? Maybe, assuming I come up with a more natural-looking rendition of a similar galaxy field.