ccdstack pixel math

[Joshua Bunn]

HI,

Say there are some stars in the colour data that are a little too saturated in terms of too much exposure. This maybe contributing to star colour loss and so can I divide the pixel values of the whole image by 2 for example to "desaturate" them thus bringing back some star colour?

thanks
Josh

[jase]

Hey Josh,

You should first try to investigate the cause of the saturated stars. Activities like combining subs using the sum algorithm will cause stars to reach saturation, especially if using a 16-bit FITS with a 65k count limit. Floating fits does not suffer from this problem. Usually however the data is automatically scaled when saving to a 16-bit format so its not a big issue but something to be aware of.

The issue with using pixel math as you suggest is that it is applying it to every pixel including that of the extended object. Thus, you inevitably need to use non-linear data stretches to pull the extend object out of the background count whilst still managing the star intensity. It doesn't get around the problem.

Using DDP is a great way to get good stellar profiles too. Other than non-linear stretches, you could also try taking short sub exposures just to bring out the stars and merging this data with the longer sub exposure master. Takes some practice but can be achieved in Photoshop with relative ease using lighten mode and keeping the extend object stars layer dim.

You may also wish to check out this post on managing stellar profiles - http://www.iceinspace.com.au/forum/showpost.php?p=666789&postcount=40

What pixel math is great for is normalising masters of narrow band data so that the applied weighting on the combine can be correctly performed. If a master has a bias i.e. overall brighter than the other masters it will dominate the scene. By subtracting the a value from this master to match the background of the other masters, you've got a much better chance of reasonable result. A 4:1:3 weighting for SII:Ha:OIII respectively is often a good starting point following normalisation.

[Joshua Bunn]

Hi Jase,

Thankyou for your help here and tips, i understand mostly what your saying.

Just a little too long on the exposures caused a little saturation - not to much, i will be posting the result within the next few days or so. I am saving data as 16 Bit float point in ccdstack.

Looking at adam Blocks tutorials, I tried using the replace missing values function to regect the blooms and replace with the less exposure data. I didnt have a wide enough dynamic range though. But, hey... its not overly saturated.

Josh

[jase]

Hi Josh,

I don't think I've used the replace missing values functionality. I've seen it work well on M42 but not star blooms. If you've got an NABG camera, then you will certainly need to watch the exposure times to get the right balance between signal and blooms? CCDStack's bloom removal functionality is extremely good especially if dithering between subs. I think a saw in another post that you have an STL11k. This is an ABG camera so you shouldn't be getting blooms. Go long exposures and let the pixel wells bleed charge to get improved signal. You can still take short exposures and experiment with a stars only layer in Photoshop. NABG or ABG, you still need to respect how the data is stretched in order to control the stellar profiles.

[Joshua Bunn]

Yeh blooms arent really a problem, im just reaching saturating. Yes i have a STL11000, I dont find blooming to be an issue but i have seen completely wiped out areas such as the core of M42 if given enough exposure time. Thanks Jase.

Josh