From: "doodlebun" <gbleser@bellsouth.net>
Date: May 19, 2008 2:58:44 PM MDT
To: Astro_IIDC@yahoogroups.com
Subject: Re: Use of Histogram expand and color balance techniques
--- In Astro_IIDC@yahoogroups.com, Milton Aupperle <milton@...> wrote:
Hi doodlebun and Gail;
On 19-May-08, at 12:25 PM, doodlebun wrote:
My wife and I use Astro IIDC with the DMK21AF04 and Astronomik
LRGB
filter wheel for imaging Saturn. My wife feels that she gets the
best
results leaving Histogram expand on to achieve proper brightness
on
all channels.. I feel that it can give Astro IIDC fits.
Actually it could cause "fits" so to speak, depending on which
type
of "Histogram Expand" you use and how high your gains are.
Astro IIDC 4 has two settings for Histogram expand, "Auto"
and "Once".
If you leave it on "Auto", then it dynamically adjusts the range
of
values for each frame it receives.
If you leave it on "Once", then it's calculates the Histogram
Expand
values from the next frame you receive and leaves it that way
until
you select "Once" again or Stop / Start the camera again.
The difference is that in "Auto" with high gains or turbulence,
the
dynamic range will change for each frame which causes the
brightness
of the image to change each time it is received. This could cause
the
"fits" or rapid frame to frame changes in brightness I think you
are
talking about.
So the "Once" method may help improve contrast and at least it's
consistent frame to frame.
My understanding is the proper way to achieve color balance is:
Take a bunch of movie frames of an out of focus G2V star using
the
monochrome camera. Start with some red frames through a red
filter,
making sure the image is not clipped. Do the same with G and B
filters, taking care to not allow the exposure to change by
keeping
all "auto" functions disabled in Astro IIDC.
If you check these star colours out at:
http://www.vendian.org/mncharity/dir3/starcolor/
you'll see that a G2V stars like our sun are not really white and
have color values #fff5f2 which is Red = 255 G = 245 B = 242 (R
100%
G 96.1% B %94.9). That makes it yellowish with slight pink tinge.
A
better star for pure white would be an F8(V) star #fff9f9 which is
Red = 255 Green = 249 Blue = 249, which is closer to "white".
Use photoshop to determine the intensities of the R,G and B
components in each image. From those intensities one derives
the "boost" value given to the dimmer G and B channels of the
tiffs
generated from stacked Saturn channels.
That sounds do able, assuming your known color star and Saturn
were
taken at near the same angle altitude above the ground and
shooting
though similar skies.
Basically you image an out of focus star of known colors (i.e. R =
100% G = 96% Blue = 95%) using the same exposure times for all 3
filters, being careful not to saturate it to 255 on probably the
red
color channel. Ideally you want to get around 55 to 65 % of the
maximum value (i.e. for 8 bit cameras 140 to 166 values and for 16
bit 36,000 to 42,600 values) for the brightest.
Then use the measured recorded R G B brightness of the stars
intensity and then compare that against the expected absolute
colors
and adjust gains as necessary to make the two match.
This method means that the RED channel of Saturn is filmed first,
as bright as possible without clipping. Then a green and blue
movie
is filmed without altering any gain, or any other setting period.
The
blue movie will be quite dim in the series. There is a strong
tendency to boost gain or exposure time, but it seems that would
throw the whole color balance scheme into confusion.
With gains I'd agree because they aren't usually linear in the
cameras.
However if you adjust only exposure time, you can also balance the
color balance that way too. For example if you determined that you
need to linearly adjust brightness by say Red = 1.00, Green = 1.05
and blue = 1.15 times to balance using your known star, then if
you
were shooting Saturn with Red at say 50 ms, then you need to use
52.5
ms Green and 57.5 ms for Blue. Also by adjusting exposure time,
your
not increasing gain which boost noise levels. Even digitally
increasing gain in post processing will increase noise levels,
where
as exposure time won't.
Use photoshop to increase the gain in each channel according to
the
test results from the G2V star. If everything is done correctly
then
you should get rings of Saturn that are actually "white" like they
are supposed to be.
And are the rings actually "white"?
Here are shots from hubble:
http://heritage.stsci.edu/2001/15/big.html
and shots from Cassini
http://apod.nasa.gov/apod/ap040723.html
http://en.wikipedia.org/wiki/Rings_of_Saturn
I've always thought of them as pale bluish to pale yellowish in
color, depending on orientation with the sun, as they are ice
reflecting light back from the sun.
And there will be no reddish bands on the planet
like I have been getting.
Umm, there are reddish and greenish bands on Saturn.
http://solarsystem.nasa.gov/multimedia/display.cfm?IM_ID=186
One thing i have done is too image terrestrial objects of known
color
and then adjust the controls visually until they balance on screen
and what I see. The ultimate means of doing it is to shoot a
standard
color chart under fixed lighting with each filter using the same
exposure time and then adjust each color channel until they
balance
out properly. Of course when the image is viewed on different
monitors with different Gammas and different OS's , it will change
so
there aren't any real absolute colors here, we can jst get "close".
Hope some of this helps..
Milton J. Aupperle
Thanks Milton,
Gail is beaming now. She has been telling me that our sun, a G2V
star, is yellow, not white. I say that everything on earth is
illuminated with our G2V star so it only makes sense that the
internet chatter is full of references to finding a G2V star for your
color balance. Yet I agree with you that white is R=G=B=255. Color
discussions can make normal people go crazy. When women consider
themselves too pale, they put some rouge on. It makes them look
better even though it's not the "real color". When we image planets
we put our images through the "anti-cosmetic" treatment.
If our planetary imaging techniques with the wavelet filter Fourier
Freakouts were applied to images of a person's face, we would
emphasize every pimple and wrinkle that we could find to bring it out
from the background. We would all be fired as photographers
specializing in portrait photography. So the rule simply
becomes "don't turn your planets into false-color clowns with details
that arise from amplification of noise rather than real planetary
features".
One wonders just how dim light is on Rhea? Here on Earth the flux
at noon is 1000 J/s/m^2. Surely you could read a newspaper through
your space helmet? Whip out a color chart on Rhea and see how much
saturation remains in the dimness. Here on Earth our software
histogram adjusters can fully illuminate a shadow scene on Pluto.
Last night we had lost 1-2 magnitudes from the sky due to smoke.
Almost as bad as the smoke in Glacier National Park in Aug 2001 that
resulted in Gail and I leaving for Calgary..and Banff. Photoshop
removed the haze in my Canadian Rockies images so well I began to
believe that all photos of ultra-clear vistas must get the Photoshop
treatment.
Gail is using Photoshop to fix the yellowing we got in last nites
Saturn images.
And before I say goodbye, do you recommend the 4.04 upgrade for us
monochrome planetary (never lunar) imagers?