RE: [ccd-newastro] Re: Help -- New to RGB
- I like to always take at least 8 darks and combine them using Russ Croman's
RC-Astro Console. But a median combine if you have cosmic ray hits will be
good enough for most.
Noise adds in from every source, including dark and bias frames. So if you
scale darks, you will have two additional noise sources: the net noise of
the combined bias frames, and the noise from scaling (scaling will never be
perfectly accurate, as the thermal current is never perfectly linear). Both
are small, but noticeable. So the ideal is to take darks of the same
duration. But I would test scaling to see if you are satisfied with it; the
difference is usually small if you follow the rule below.
Ron's Rule for darks: take at least as many dark frames as light frames and
Corollary #1: Take as many bias as darks if scaling.
Corollary #2: The darks for scaling must be as long as the longest light
frame. Due to the nonlinearity of dark current, I don't recommend taking
extra-long darks when scaling.
The New CCD Astronomy
From: ralldredge2001 [mailto:ralldredge@...]
Sent: Sunday, May 02, 2004 11:12 AM
Subject: [ccd-newastro] Re: Help -- New to RGB
Thanks for your advice. It sounds like modifying exposure length is
the way to go. Do you just take a humongous number of darks -- or
is it possible to scale the darks using bias frames? And how well
does this work?
--- In email@example.com, "Wodaski Yahoo" <yahoo@n...>
> If you want to do it numerically, you would need to do it in MaxImDL where
> you can see the numbers. The problem is that it involvesiteration. First
> you clean up backgrounds, then scale the data, then you clean upthe
> backgrounds because the scaling affects the background levels aswell as the
> foreground, and so on. I find it simpler to do this visually inPhotoshop.
> In my classes, I recommend against shooting color at the same
> because of some problems that can occur. I'll try my best toexplain without
> the advantage of illustrations. <G>times
> There are two things to aim for when imaging color: using exposure
> that will make sure of accurate color in the bright objects, andhaving
> enough total image time for each color to provide similar noiselevels in
> the dimmest areas (dark zone).both the
> When you scale an image, you are scaling everything in the image -
> signal and the noise. If you increase the signal, you must alsoincrease the
> noise (the signal to noise ratio remains constant when you scalethe data).
> If you scale, for example, the blue image, then the background ofthe blue
> image will be noisier that when it started out.overall
> This is why I recommend against scaling; it creates problems for
> color balance. Your stars and galaxy cores may have accurate colorafter
> scaling, but the dimmer portions of the image suffer from colornoise and
> this prevents you from displaying the Dim data effectively.up
> The normalization technique outlined in the first book will clean
> backgrounds. If you then scale, the backgrounds will be differentagain, and
> must be renormalized. This can also be done numerically in MaxImusing the
> same technique. But renormalizing may change the color balance,and require
> another round of scaling....you get the idea. You continue untilthe changes
> are small enough not to matter, iterating toward a good balance inachieve -
> foreground and background. Thus simple ratios are difficult to
> hence my recommendation to do the balancing in Photoshop.the
> It's not in the first draft of the book, but consider displaying
> Histogram palette, and turn on display of the separate histogramsfor all
> channels. That will give you the ability to get numbers - pass themouse
> over the Histogram palette for each color, and you will get anumeric
> display of where you are in the histogram. This might help youintegrate
> both techniques at the same time. The next draft of the colorbalance
> chapter will include details on this technique.is
> Ron Wodaski
> The New CCD Astronomy
> -----Original Message-----
> From: ralldredge2001 [mailto:ralldredge@p...]
> Sent: Saturday, May 01, 2004 8:02 PM
> To: firstname.lastname@example.org
> Subject: [ccd-newastro] Re: Help -- New to RGB
> Actually, I have bought your new book and have downloaded your
> current chapters. Your chapter on color balance does give me the
> basics ---- But, I was hoping for a more mathematical way to do to
> first color combine, especially when a typical Don Goldman ratio
> something like .92, 1, 1.71. This means, I think, that I must adda
> lot of blue. Adjusting the histograms using the mid-point sliderby
> eye can give me a pleasing picture. But it is clear to me that myin
> eye is not yours.
> --- In email@example.com, "Wodaski Yahoo" <yahoo@n...>
> > You need my new book. There is a full chapter on color balancing
> > Photoshop. The book is incomplete, but you can pre-order it andcombine
> get access
> > to the content online as I write it. Please see:
> > http://www.newastro.com/ipb
> > Ron Wodaski
> > The New CCD Astronomy
> > http://www.newastro.com/ipb
> > -----Original Message-----
> > From: ralldredge2001 [mailto:ralldredge@p...]
> > Sent: Saturday, May 01, 2004 11:18 AM
> > To: firstname.lastname@example.org
> > Subject: [ccd-newastro] Help -- New to RGB
> > I am new to RGB and have just taken some images. I assumed that
> > equal exposure times would be the way to go, as I could combine
> > in CCDSoft and would only need one set of darks.
> > Sure enough I could combine them in CCDSoft, but they were
> > and the core of all of them is burnt out when using any of the
> > histogram settings which seem available through the color
> > command. Frankly trying to maintain color balance while usingthe
> > set of curves in CCDSoft exceeded my abilities.
> > What I really want to do is use Photoshop CS. I now can color
> > combine using 16 bits, and do curves and levels. However, I
> > figure out how to add the RGB weighting. I can wing it using
> > color balance commands, but it seems a shame that Don Goldman'splaces
> > combine program will give me ratios to two or three decimal
> > and I just have to guess in Photoshop. Am I missing something?Yahoo! Groups Links
> > there a way to do this?
> > Yahoo! Groups Links
> > __________ NOD32 1.745 (20040501) Information __________
> > This message was checked by NOD32 antivirus system.
> > http://www.nod32.com
> Yahoo! Groups Links
> __________ NOD32 1.746 (20040501) Information __________
> This message was checked by NOD32 antivirus system.
__________ NOD32 1.747 (20040502) Information __________
This message was checked by NOD32 antivirus system.