|
How many calibration frames (darks and flats)?
Stand by for further coverage of this later, but here are some preliminary results.
With the help of some other astrophotography experts
on
the Cloudy Nights forum, I've been doing mathematical analysis
of the way deep-sky images are calibrated and stacked. We normally take many exposures of
the object ("lights"), matching dark frames to counteract fixed-pattern noise in the sensor ("darks"),
images of a flat field to counteract pixel non-uniformity and dust specks ("flats"),
and "flat dark" or "bias" frames to subtract out the nonzero baseline in the flats.
How many of each should we take? Preliminary results:
Lights: As many as possible. My current practice is to take 20 to 120 2-minute exposures.
The biggest source of noise in the finished image is shot noise (Poisson noise) in the lights,
and if you take N lights instead of one, you have 1/sqrt(N) as much of this noise.
Darks: Need not be as numerous as the lights, because they do not have shot noise from the image,
although they still have read noise. Their purpose is to correct hot pixels and fixed-pattern
leakage. Some of the newest sensors may not need them. My practice will be to take 10 to 20.
Flats: Need not be numerous because any noise in them is small compared to the signal.
Expose them as generously as you can without actually bumping into the sensor's maximum value,
so that their read noise will be as small a fraction as possible of the total output.
Generous exposure helps more than taking a lot of flats. My practice is to take 10, which may
be more than are needed, but they're quick to take.
Flat darks or bias frames: A few are sufficient; I currently take 10.
The big discovery is that dithering is VERY important. Dithering means not aiming the camera
at exactly the same point in the sky for all the exposures. That way, when frames are combined, lined
up on the stars, any noise that is in fixed positions on the frame will be spread around and averaged
out.
And where does this noise come from? Residual random noise in the master dark and master flat.
Each of those has a small amount of read noise that couldn't be eliminated, and it's in a fixed position
because the same master dark and flat are applied to all frames. So the frames need to be dithered
on the stars, to spread this residual noise around. The benefit can be dramatic.
Even a little dithering helps a lot. It can be sufficient to spread the whole session across just
four or five positions. So when I'm using a star tracker that can't dither under computer control,
I'm going to simply stop it and re-aim it periodically, shifting randomly a small fraction of the
field of view.
Some time back, Robin Glover demonstrated that if the exposures are reasonably generous
(so that the sky background is appreciably lighter than black), the read noise (random noise)
in the sensor is surprisingly unimportant; we can put the brakes on our headlong rush toward
lower-noise sensors, although I still want one because they tend to be better in other ways too.
That conclusion holds up through all my analyses.
Note however that his recommended exposures pertain to faint objects and assume that, in the
finished picture, the sky background will be medium-dark gray.
It is quite legitimate to use shorter exposures if you are photographing a relatively bright
object (such as M42) and are going to render the sky darker.
Permanent link to this entry
News updates
Sadly, for several reasons, we're not going to get to go to Indiana
to see Monday's solar eclipse, as we had planned. I will observe it from here, where
it will be a deep partial eclipse, and hope that the grandchildren have a good trip to
totality, which passes near their home.
Not only is Melody not quite up to a long road trip yet, yesterday I had some
urgent dental work done. I was sent to Tucker, Georgia (Melody's childhood
home town) when the dental surgeon in Athens suddenly had to cancel.
So I was right across the street from where she went to 8th grade, and down
the street from where she went to church, and even got to see (momentarily)
the public library. And today I'm eating applesauce for breakfast.
Hoping to be back to normal soon, but glad not to be driving to Indiana today.
Meanwhile, the State of Georgia has stopped publishing the blue COVID graphs
that I've quoted so often. I'd say, "The war is over," except that I had
COVID just two months ago.
Permanent link to this entry
"Thou shalt not bear false witness against thy neighbor"
Another aspect of weaponized stupidity
I am finding that many of my fellow Christians are very poorly instructed about
this one of the Ten Commandments.
In fact, they often feel they are doing very righteous things when they violate it.
The commandment says not to make false accusations.
I take that to include not spreading accusations of wrongdoing without
good evidence and a good reason to spread them.
Even when the one accused is an organization, or a person known to have other
serious faults.
If you're saying someone did wrong, this commandment applies to you.
I know that sometimes, someone else's wrongdoing is a matter
of public concern. If so, I think you still have a duty to be honest about the
evidence, about whether the wrong is proven or just suspected.
"I heard" or "somebody said" is not good enough.
Nor is "my best friend shared this on Facebook" where "this" is something from a
perfect stranger.
I have had to rebuke some of my fellow Christians rather strongly for spreading
the hoax that "Facebook has banned the Lord's Prayer." Facebook has not done that,
and if you check, you'll find no evidence that it has. There is no controversy.
Just gossip.
But people spread it the moment they hear it from a total stranger.
And they feel so righteous when they spread this malicious lie!
Maybe because they're blaming an organization rather than an individual;
maybe because they think they're supposed to hate secular organizations,
even when relying on them; or maybe because they're just foolish.
There's a deeper layer here. What I am realizing is that some people don't
even know what truth is.
I take it for granted that truth is facts; truth exists in the real world;
truth is what you can confirm by looking at the real world.
For some, though, truth is just words running around in your head,
things you're supposed to say.
For some, truth is just emotions. If you "feel" like believing or saying
something, you believe it and say it. To such people, logic and evidence don't matter.
It is sad, but such people are surprisingly numerous.
And for some, truth is a matter of taking sides, like choosing to cheer for the Gators or
the Bulldogs; and then you're supposed to say what your side says, or what
seems to show loyalty to your side.
I think that with the Lord's Prayer Hoax we're seeing the last of these.
People think that to be loyal Christians, they are supposed to believe and spread
any accusation of anybody persecuting Christians, whether or not it is true.
If the Book of Proverbs is still in your Bible — if you have any idea
what godly character is — this isn't it.
Permanent link to this entry
|
