5
« on: 2012 August 31 06:36:23 »
Q) I assume it's best worked on linear data, before stretching? If so, how does one get the nebulosity the same intensity in both the R and Ha images? My Ha shots were 4 mins each at 1x1, while the the R was 30 secs at 2x2. Is it just done by eye and HistoTrans'n, or is there a definitive way?
A) Yes, linear data. Use Harry's video formula, you don't need to linear fit the data. The thing is what you need to do is to cancel the star flux so only the real Ha is added to the image. If you do this well, stars will be unaffected, and only the nebula SNR will be raised.
The bad news is that you have to experiment with the bandwidth, until you reach the best results.
Q) Vicent talks about iterations after combining the narrowband image (N) back with the noise reduced continuum image (Cnr) to get a new broadband R image (Bnew). Not clear, but does he mean using Bnew, divide by orig N to get a new Cnr2? Or use the Bnew as the new N, so C=B/Bnew? Sorry, just read that back and it's a mess. Hope it makes sense.
A) Just use Harry's formula, no brainer here.
Q) All the posts about this HaRGB combination just talk about the Ha and RGB. Haven't seen too much mentioned about L. Does that get added at the end, after Ha introduced into the R? Or when reference is made to the RGB, is that actually an LRGB image being referred to? I don't think so, because as I understand it, L is best combined with RGB after a bit of stretching?
A) Two options here:
1) You didn't capture L. Just put your Ha image as the L. You'll end with a Ha(Ha+R)GB.
2) You did capture L. You need to combine somehow L and Ha. Use pixelmath to reach the best results here. I recommend L + (Ha * constant) here.
Q) Some of the formulae talk about the filter bandwidths. I'm using 12nm Astronomik for Ha (at the moment, will shortly be going to Astrodons), but I have Astrodon Gen 2 E LRGB filters. There is heaps of info on narrowband bandpasses, but I had to use this chart to try to guess the Astrodon RGB bandpasses. Not sure whether to use the bandwidth at 50% or 100%? At 50% the R is about 65nm, but at 100% it's around 50nm. Any comments on the importance or accuracy of this? Am I overthinking it?
A) As I commented earlier, just experiment with the data and find the best balue. There is a lot of variables here to guess the value without experimenting. I use 3 and 50-60.
Q) Similar to my question above about how to match the star brightness for the second half of Vicent's Method. By eye, or definitive method?
A) I did try to build a script for that, but it doens't work. Better by eye.
Q) Vicent's Method seems to work with division with that formula C=B/N, but some of the pixel maths seems to use subtraction instead, including Harry's video. So which is it? Please don't say both, I hate it when people say both
A) Just use Harry's formula, no brainer here.
Q) Harry's video has 2 pixel math process icons with formulae:
((Ha*100)-(R*12))/(100-12) - the 100 is R bandpass value, 12 is Ha bandpass value. This gives a new image, which as mentioned above appears to be the continuum image via subtraction rather than division.
$T+(ha-Med(ha))x4 - ha is the continuum image from above, 4 is the "boosting factor" from Vicent's Method.
Any tips on modifying the above to suit different binning and exposures? I saw one post by Ionnis(?) that did seem to take these into account, but for the binning it wasn't clear on whether it should be multiplied by 2 for 2x2 binning, or by 4 for 2x2 binning.
A) Use the same formulas, the only variables are:
FWHM of the Ha filter -> 12, fixed
FWHM of the R filter -> guess it!
Boost factor -> the one that better catches your eye, if set too high, it will increase noise
Q) Might be getting ahead of myself considering the enormity of above post, but I assume similar theory could be adopted for OIII in the G and B channels?
A) Yes. It's a super simple process, just find the better values and voila, you're done combining the data.
Show Posts
