Narrowband Imaging questions and answers and tips

[Carole]

Ivor suggested that as we are both new at this we start up a thread with things we have learned, things we need advice with. So here goes.

Q:  How do you prevent getting red stars when combining Ha (red) with other filters?

Tip: I managed to prevent the hot pixels being stacked by altering the cosmetic settings in DSS and stacking kappa Sigma stacking, but is there a better way?

Carole

[Ivor]

Thinking about this perhaps the question should worded more broadly:

What methods could be used to bring back stars to their true colour and remove any unwanted halos produced from a strong NB channel?

Is it worth making the channel data available so others can experiment?

Some possible things to try:

1)   In PS take a couple of colour points from the halo, identify the weaker channel(s) and using curves boost the channels at these specific points to bring it closer to true colour.
2)   Prior to PS when stretching the channel increase the peak value of the weaker channels to create a better balance in PS.

[Mac]

Q:  How do you prevent getting red stars when combining Ha (red) with other filters?

Im assuming that you are combining all narrow band images here and not HA onto a one shot colour

although the answer is probably the same,

If you are getting bloated red stars when you add the HA image to another filtered image, then this sugests to me the following.
The image that you have just added is either out of focus compared to the other image that you are adding to,

or

The image is over saturated. i.e the exposure is longer and the light has spread out to the next pixels.

If you take a single image either one shot colour or NB, then the stars should produce a point source,
If the image is out of focus (and we are not talking a great deal), then that point source is now over a greater distance.

For arguments sake, instead of the star covering 9 pixels, it is now covering 25 (ever so slightly out of focus or a longer duration of image),

Now if this image is added to another image that is in focus, then the two 9 pixels will overlap, but you now have a 1 pixel halo around the other star.


***                           ***
***                           ***
***                           ***

overlapped

***
***
***

slightly out of focus

*****                           ***
*****                           ***
*****                           ***
***** 
***** 

overlapped
***** 
*****
*****
*****
***** 

the problem here is that if you take a 30 second image of Ha, Hb, Oiii, Sii  ect ect,
then the amount of light that is recorded will be different due to the nature of the filters.

You might find that you need to take 3 times the length of exposure in Hb, compared to Ha,
also if your filters are not par-focal, then you need to refocus each image when you change the filter.

To get your timings for each filter i would do the following.
Take a sheet of white paper, illuminate it with white light, and take an image of the sheet of paper.
Draw a shape on it and focus on the shape
(do it in daylight if you must, but you need to keep the illumination even.)

using each of your filters in turn, BUT make sure that the exposure is identical for each of the filters take an image,

then using something like photoshop measure the grey value of the image produced.
ideally all of them should be identical as the exposures are identical, but you will find that they vary.

Using the highest value as 1. calculate the difference as a ratio and use this to change the timings.

i.e. if the Ha gives you a value of 128 and the Hb is only 64, then you need to expose the Hb for twice as long to get the same exposure.
once you have sorted out these timings, try again and compare the results

Now any images you take using the NB filters, using the adjusted timings, should have exactly the same star sizes
for any given exposures.

Hope this makes sense

Mac.

[Rocket Pooch]

Gosh!

[RobertM]

Makes sense to me.  In your case Carole, another reason for larger Ha star images is that your Ha filter is 12nm whereas the OIII and SII are much narrower (7/8 nm).  That will exacerbate the problems Mac has mentioned below.  Usually Ha is significantly brighter so it makes sense to have a narrower bandwidth there (if you can) but not wider.  By using a narrower Baader Ha filter like your OIII/SII you will also benefit from significantly better signal to noise on those Ha exposures.

Robert

[Carole]

Makes sense Mac thanks - focus was OK with all filters Mac I refocussed for each one, (except the Baader ones were shown to be parfocal and didn't need to change anything as I could see focus was the same) and all subs were the same length, except I did less of them for the Ha than Oiii and Sii.  

Robert, your comment about the Ha filter being a different bandwidth is what I thought was the reason, and made such a comment on another forum, only to be contradicted by two experienced imagers.

I am planning on getting a 7nm Ha filter, but having just shelled out for the camera, filter wheel and Oii and Sii filters, I thought I would use the 12nm Ha that I already had for now and see how it would come out and my conclusion was that I needed a 7nm filter.

so it makes sense to have a narrower bandwidth there (if you can)

Ah, that might explain why they sell the NB Baader filters as Ha 7nm, Oiii 8.5 & Sii 8nm. 

So in basic summary I have to do a combination of 7nm Ha filter and possibly shorter subs than the other filters to correct this then?

Carole

[Ivor]

Mac a great explanation thanks, what you describe appears to be an alternative to G2V star calibration I guess.

One thing I'm not sure about, I see this working for an RGB image where you are striving for a true colour image, in the case of NB where you might want to extend the sub time to get the extra detail for SII and OIII, bringing HA up to be in ratio with the others could lead to over exposure.

If an image is approached in this way what are the best methods in post processing for bringing colour balance back to the stars?

[MarkS]

I have a nasty suspicion that it isn't that simple.

It might be the case that you need one colour balance to give stars the "right" colour and then a differenct colour balance to get the areas of nebulosity to "look" aesthetically pleasing.

Mark

[Mac]

So in basic summary I have to do a combination of 7nm Ha filter and possibly shorter subs than the other filters to correct this then?

in theory yes.

If you Ha was 180 secs, then your Hb,might be 240,

Now, here's the killer question.

If you were shooting narrow band on a dslr (modified), then you can take each separate image and
post adjust the exposure using your RAW software (whatever you use) to compensate for the halos.

So if you were able to get a combined Ha, Hb Oiii narrowband filter. 

http://www.meade.com/catalog/meade_4000/meade_series_4000_nebular_filters.htm 

then you could one shot colour using the dslr, and then extract the separate RGB layers, which would be narrow band images.
and if you then have halos, by changing the raw exposure you can adjust out the halo and then reuse the separate adjusted RGB layer

Mac

[RobertM]

Carole,

One point of note is that these narrow band filters attempt to isolate the spectral emissions of some elements and the wavelengths are exact values (though they can be doppler shifted).  By choosing a wider bandwidth filter you are capturing other light that is not for example at the frequency of Ha.  This by definition means that you are capturing light that you don't need or want i.e. additional starlight/light pollution etc.  Additional light from stars would have a bloating effect, making the star image larger.  If you look on decent images in Ha light you will notice that the stars are quite small.  On extremely narrow band filters this can cause problems with the size of stars from different filters not being of the same size.  You might notice this as a different coloured halo  around the brighter stars in some NB pallette images (even from the HST).  Ideally I would have ultra narrow band filters but they do cost a fortune

Robert

[Carole]

Thanks Robert and Mac.

Carole

[Mac]

One point of note is that these narrow band filters attempt to isolate the spectral emissions of some elements and the wavelengths are exact values (though they can be doppler shifted)

Yes and No.
Yes in the fact that you are completely correct, in that the further the galaxy the greater the distance in the red shift.

No in the fact that the distances that you are photographing (with the exception of galaxies, every thing is in our galaxy) will have none or a red shift that is so small its insignificant. (i'll do the maths later). Or they could even have a blue shift. Dont forget when they travel towards us the llight is blushifted,
thats how they measure the rotation of galaxies.

As for the 7nm bandwith, this is the bandwith measures at the FWHM level.



or in old money, the 3dB level.

Mac

[MarkS]

Interesting points Mac.

What's the max red/blue shift in the local group of galaxies?  i.e. how narrow a passband is acceptable when imaging the H-alpha bits of M41, M33 etc?

[RobertM]

Mac, I only put the doppler shift in to qualify the 'exact values' and not for imaging purposes otherwise someone else might have done that.

Thanks for filling in the background though.

[MarkS]

The blue shift of the Andromeda Galaxy in terms of wavelength is around 0.2 nm at the H-alpha wavelength according to this:
http://en.wikibooks.org/wiki/A-level_Physics_(Advancing_Physics)/Doppler_Effect/Worked_Solutions

So it won't be a problem even for a 3nm bandwidth filter.

The blue shift is small enough that we don't need to consider relativistic effects in the Doppler calculations.

Mark