Blobby Stars

[MarkS]

I've been suffering from very blobby stars in many of my images.  Yesterday I sat down and decided to see if I could do anything it.  So I've completely reprocessed one of my worst affected images.  To do the comparison it's best to download both and "flick" between them. See what you think.

Here they are:
http://www.markshelley.co.uk/Astronomy/2010/ngc6888_09072010_crop.jpg
http://www.markshelley.co.uk/Astronomy/2010/ngc6888_09072010_cropv2.jpg

Mark

[Carole]

I guess version 2 is less blobby, but version 1 is prettier.

Carole

[mickw]

V2 looks sharper but you've got haloes round the stars and the colour is shifted.
If it wasn't for the haloes V2 looks better

[Fay]

 prefer number 1

[Carole]

prefer number 1

Yes so do I.

Carole

[JohnP]

Ummm.. been flicking between these two for a while. Stars are obviously smaller on 2nd but I think you have over done it - they look too small.. also looks like you have clipped the whites on 2nd (cores are burnt out on stars - not sure if this is a consequence of making them smaller). Also colours have changes a bit - I think I prefer colouration of the original. If I had to pick I would probably say original but if you could come up with something in between the two I think this would be best. I always find NGC6888 a difficult object to process - my images always look soft but I think this is a result of it being shrounded in nebulosity..

Anyway its a great image - John.

[MarkS]

Interesting comments.

So it seems that, in general, folk object to the dark haloes that appear around the core? 

I think it depends on the original star brightness.  I much prefer the medium brightness stars in the second image because they appear tighter.  However, I agree the effect is too strong on the brightest stars - the dark halo has the effect of "disconnecting" the diffraction spikes from the core.

I'm starting off with an HDR image and I'm playing around with the "transfer function" that creates the final image from this original.  I'll keep playing ...

Mark

[mickw]

Just had another look and I was wrong with my previous comment, the colour isn't shifted.  Because the stars are tighter the colours are less pronounced particularly the cyan (with my monitor and eyes anyway)

[RobertM]

The bright stars look too too burnt out in the second to me, they also look burnt out in the first but not so much.  Overall though I much prefer the second, other than the issue with bright stars the rest look much tighter and there's more nebulous detail present.

If you could keep the bright stars that tight and also soften up their edges then the second would do it for me.

Hope that helps
Robert

[MarkS]

As a matter of interest, I already used this technique for tightening up the stars on the M13 image I posted last week:
http://forum.orpington-astronomy.org.uk/index.php?topic=7551

[Fay]

Can you explain that technique please Mark?

I did an M13 but cannot get the central stars tight enough

[The Thing]

Star tightening.

Can you explain that technique please Mark?

I did an M13 but cannot get the central stars tight enough

Star tightening. Yes my M13 could use some of that as well!

[mickw]

Interesting, the haloes are on the M13 as well, but not so pronounced.

As the backgrounds (nebulosity/black) are different could that be affecting the process ?  sharpening the stars by introducing a black point rather than matching the background ?

[MarkS]

Mick,

The method will not create dark haloes around stars on a dark background but it can create dark haloes around stars whose background is nebulosity.

In a raw stack, the bright peak of a star must be scaled back a lot for the final image - you might do this by applying an arcsinh or log transform - during such a transform, faint parts of the image are left alone whilst bright parts are scaled back.  The method I'm experimenting with works by increasing the scaling of the pixels in the neighbourhood of any very bright areas such as the centre of stars.  If this method is applied too severely, the increased scaling of the neighbourhood pixels will create the dark halo around the star.

The effect of this method is to tighten stars and to increase detail/contrast in the brighter areas of the image such as galaxy cores and centres of globular clusters.

Very briefly the method does this:
1) Divide the raw stack by the arcsinh (or log) transformed stack.
2) Apply a small amount of gaussian blur to the result (e.g. 0.4 pixels) and save it.
3) Divide the raw stack by the frame you just saved.

Mark