►
Description
Astro-processing: What I learned in the past year
Hy Murveit
In the past couple years we've had excellent talks to the SJAA SIG, and many were concerned with improving our techniques for process astro-images. There have also been significant advances in the software available to process those images. Because of all that (and the rain), I was inspired to review the past talks and reprocess many of my images. I found that, to my eye, I was able to significantly improve most of them, starting from the pre-processed images and scrapping most of the old work. In addition, the processing took me much less time than I had spent in the past (probably a couple hours per image), and produced less frustration as well.
In this talk I'll go over the basic workflow I used, heavily borrowing from recent talks. My images are almost all taken from my home in Silicon Valley, using 10-30 hours of mostly LRGB images, and mostly from a 10" reflector. I'll present the WBPP, synthetic L, blurX/noiseX/starX, GHS, Photoshop/Lightroom workflow I used, and show a few full images being processed.
A pretty complete before/after photo album demonstrating the improved images can be found at
https://photos.app.goo.gl/yYx6U5rZ25zGDjcFA
A
Welcome
to
the
February
2023
sjaa
Imaging
Sig
meeting
and
tonight
I'm
going
to
speak
hi,
mervite
I'm,
going
to
speak
about
things,
I
learned
in
the
past
year.
We've
had
several
meetings
of
you
know,
helping
teaching
us
at
least
you
know
four
by
Francesco
Alex
has
given
a
talk
and
there
may
be
others
I've.
Oh
Dave
Payne
gave
two
presentations
and
there's
also
so
I've
learned
a
lot
from
those
talks
and
hopefully
I,
haven't
missed.
A
A
I
did
the
reprocessing
and
I
found
that
my
newly
processed
images
were
a
lot
nicer
to
my
eye
than
than
the
ones
that
I
had
been
pretty
happy
with
previously
much
to
my
surprise,
honestly
and
so
I
wound
up
reprocessing
one
and
then
I
wound
up
reprocessing,
this
second
and
then
the
third
and
so
on
and
so
forth.
I
wound
up
spending
like
all
of
January.
You
know
reprocessing
at
least
20
images
and
he
sent
that
out
and
I
encourage
you
in
the
first.
You
know
five
minutes
of
this
talk.
A
A
You
know
hopefully
not
too
much
time
on
each
image
but
I'm
going
to
go
more
detail
in
them,
but
here
I
think
I
have
20
images
in
this
slideshow.
So
if
you
click
on
I'm,
going
to
click
on
the
first
image
and
then
use
the
arrow
key
to
advance
from
one
image
to
another
and
you'll
in
all
cases,
the
first
image
of
a
particular
object-
you
see
is
the
old
version
you
know
done
without
blur.
X
done
without
the
general
hyperbolic
stretch
done
without
the.
A
B
A
Oh
done
without
weighted
by
badge,
processing
spcc
was
the
thing
I
forgot
anyway,
and
then
maybe
just
the
things
I've
learned
from
these
talks
in
the
past
year
and
then
the
second
one
is
my.
You
know
revised
image
that
I
did
in
January,
and
so
here
you
go.
Here's
M13,
you
know
a
nice
big
cluster
by
the
way,
unless
I
say
it,
otherwise,
they're
all
shot
with
my
10
inch
RC.
A
So
the
the
the
width
of
the
image
is
about
a
half
a
degree,
so
you
know
a
pretty
long
setup,
not
planetary,
but
longer
than
you
know,
a
100
or
130
refractor
and
anyway
you
could
see
here.
Here's
M13
which
I
used
to
be
happy
with,
but
a
much
sharper
version
of
M13
after
reprocessing
and
not
only
sharper
with
the
Stars
being
tinier,
but
little
galaxies
like
the
one
in
the
upper
right.
A
A
This
is
I
believe
the
only
image
that
has
any
dark
sky
data,
Francesco
and
I
went
down
to
Pinnacles
like
in
2019
one
weekend
or
and
and
this
was
an
image
that
I
shot
part
of
the
data
came
from
that
and
then
more
came
from
home.
All
the
rest
of
the
images.
Our
data
that
came
from
home
I
live
behind
silica
behind
Stanford,
so
it's
you
know
better
Skies
than
Downtown
San,
Jose
or
downtown
San
Francisco,
but
far
from
Dark
Skies.
A
You
can't
see
the
Milky
Way
or
anything
like
that
from
my
house.
It's
I
think
it's
five
is
the
designation,
but
anyway,
so
that's
the
old
version
of
triangulum
and
that's
the
new
version
of
triangulum
just
much
more
definition
on
the
Galaxy.
You
know
again
much
better
stretched,
I
think
certainly
much
tighter
Stars,
nice,
color,
I
believe
so
on
and
so
forth.
A
You
can
agree
or
disagree,
but
that
that's
the
way
I
see
it
and
again
here's
Sombrero
that
I
was
proud
of
and
then,
in
my
humble
opinion,
a
much
better
Sombrero
taken.
You
know
all
these
things
are
the
same.
Data
data
has
not
changed
so
the
same
fits
files.
In
many
cases,
I
didn't
redo
the
pre-processing
in
some
cases
I
did
and
but
there
you
go
as
a
few
more.
This
is
an
image
I
actually
shot
pretty
recently
at
the
end
of
2022
and
the
processed.
A
Again.
Arp
273
same
idea,
I
forget
when
this
one's
name
is,
but
it's
it's
a
very
nice
little
Galaxy
I'll
I'll
stop
with
this
final.
One
I
believe
is
this:
is
this
cocoon
I
forget?
But
anyway
you
could
see
that
the
cloud
here
is,
you
know,
there's
more
detail
about
it.
This
object
here,
little
blue,
nebula,
I.
Guess
a
reflection
nebula
in
the
lower
left
isn't.
A
Even
if
you
go
back
to
the
first
version,
it's
hard
to
even
see
I
guess
it's
this
sort
of
area
here,
I
didn't
do
a
good
job
of
finding
it
and
all
of
a
sudden
it
popped
out.
I
wasn't
looking
for
it.
I
didn't
process
it
in
order
to
pop
that
out,
but
poof
I
think
the
ghs
scheme
helps
a
lot
with
all
that.
So,
as
the
you
know,
that's
my
introduction,
my
motivation
for
this.
So
what
what
I'm
gonna
do
is
go
through.
A
A
So,
as
I
said
before,
there
are
almost
all,
except
for
triangulum,
maybe
only
except
for
triangulum.
It's
noted
in
the
in
the
Google
photos
album
where
things
are
shot,
but
almost
all
are
shot
from
home.
Most
of
them
not
all
of
them
are
lrgb
some
there's
a
few
rgbs
and
then
sometimes
I
added
h,
a
data
there's
no
narrow
band.
A
You
know
O3
and
S2
stuff,
it's
just
they're
all
either
RGB
plus
l
or
plus
L
and
plus
RHA
or
just
anyway,
you
get
the
idea
more
and
more
I'm
shooting
long.
You
know
lots
of
data
because
I
have
a
pretty
stable
setup
at
home,
more
or
less
I
know
how
to
use
my
imaging
software
I'm,
one
of
the
developers
of
k-stars
I
use,
K
Stars.
It's
pretty
stable
for
me,
and
so
you
know
if
the
Knight
has
any
promise
at
all.
A
I
will
just
go
outside
uncover
my
telescope
I
leave
it
there
covered
typically
just
uncover
it
go
back
inside
and
either
if
I've
already
imaged
an
object
more
or
less.
It
takes
me
a
minute
to
bring
things
up
and
hit
start
again
and
it'll
image
or,
if
I
haven't
it
may
take.
You
know
half
hour
to
set
up
the
image
or
whatever,
but
it's
pretty
easy
to
image.
So
I
tend
to
image
on.
A
A
A
That's
some
people
buy
into
that
some
people
don't
next
week
next
month,
I
should
say:
Alex
is
going
to
give
a
presentation
where
he
discusses,
where
maybe
shooting
L
isn't
worth
it
and
can't
wait
to
hear
about
that.
But
for
now
on
this
presentation,
that's
what
I
did,
and
it
makes
sense
to
me
that
you
get
more
photons.
That
way.
I
I
will
say
that
my
Luminosity
filter,
my
L
filter,
is
not
one
that
has
a
lot
of
notches
out.
A
A
Don't
know
when
I
have
do
problems,
it's
really
obvious.
There's
an
M82
sub!
That's
actually
a
keeper,
but
I
didn't
prepare
this.
For
this
talk
but
I'm
sure
when
I
went
blank
through
my
M82
data,
there
were
things
where
you
could.
Barely
you
know
you
had
similar.
You
know
you
just
didn't
see
the
M82
right,
so
I
wouldn't
have
kept
that
I
probably
threw
out
I,
don't
know
10
of
the
data
through
blink
I.
Don't
know
if
that's
a
good
number
but
something
you
know
it
wasn't
50
and
it
wasn't.
A
A
So
this
top
control
on
blink,
which
tells
blink
to
you,
know
so
to
basically
set
the
screen
transfer
function
so
that
you
can
see
each
image
well
and
that's
what
I
do
and
that's
the
manual
obviously
a
manual
thing
and
then
move
on
to
weighted
batch,
pre-processing,
I
didn't
used
to
use
I
heard
a
lot
of
people
talk
about
wbp
and
I.
Don't
know
I
felt
like
I
could
do
a
better
job.
Pre-Processing
data
on
my
own,
using
the
kind
of
techniques
I
learned
you
know
year.
A
You
know
three
years
ago
when
I
was
starting
out.
I
found
I
was
wrong,
that
this
does
a
great
job
and
it's
totally
automated,
and
it
saves
me
a
lot
of
time
so
I'm
a
convert,
I'm,
that's
what
I
do
and,
as
an
example,
I
have
the
statistics
on
the
left
here
for
a
project.
I
recently
did
on
M82
I
didn't
bring
up
a
picture
of
hima2a.
It
was
a
object.
I
was
happy
with
the
results,
for
it
was
an
lrgbha
object
and
I
shot
a
lot
of
it.
I
just
I
I
I.
A
Just
there
were
a
lot
of
nights
in
December
or
I,
think
it
was
December
and
January
where
I
don't
know.
I
just
turned
on
the
the
scope,
and
when
that
was
up,
I
took
pictures
of
it.
It
was
probably
up
several
hours
a
night
and
I
just
kept
on
shooting
and
wound
up
collecting
46
hours.
As
you
see
here,
this
is
46
hours
post
blank.
So
let's
say
I
collected
for
the
sake
of
argument.
A
50
hours
and
I
blinked
out
the
you
know,
four
of
them
and
I
wound
up
with
46
hours
of
lrgb,
which
is,
as
you
can
see
here,
651
two
minute,
L
exposures,
and
then
you
know
on
the
order
of
a
hundred
rgbha
exposures.
The
reason
they're
not
the
same,
because
I
do
the
the
software
tries
to
get
the
same
number
of
RGB
and
has
is,
is
totally
damaged.
Data
at
least
for
the
RGB.
Maybe
the
AHA
was
a
little
different.
A
A
For
weighted
batch
processing
to
to
process
this
data,
but
that's
okay,
I
sleep
overnight
and
it
works
overnight,
and
so
as
long
as
you
learn
how
to
set
your
laptop
to
you
know
not
turn
itself
off
when
you're
idle
for
a
half
hour,
you
can
run,
you
can
run
wbp
and
it
will.
You
know
you
wake
up
in
the
morning
and
there's
or
okay.
Maybe
you
gotta
wait
for
your
shower
and
breakfast,
but
then
at
some
point
soon
there's
there
are
your
pre-processed
images,
which
is
great.
A
In
this
case
it
took
200
gigabytes
ballpark
of
disk
space.
To
do
this,
so
not
only
it
takes
10
hours
but
200
gigabytes
for
this
amount
of
data.
It's
a
lot
of
data,
admittedly,
but
that
but
then
there
you
go
and
really
it's
very
little
setup
time
I'm
going
to
show
and
once
I
switch
over
to
pixinsight
the
setup
for
that,
but
not
obviously
let
things
like
that
run
so
anyway,
after
wpp
I
have
you
know
four
or
five
four
or
five
Master.
You
know
Channel,
monochrome,
images
and
I.
A
Don't
exactly
use
the
Luminosity
image
straight
as
it
is
I
make
a
synthetic
L,
Francesco
and
I
work
this
steam
out,
you
might
think
mostly
Francesco,
and
this
game
basically
pulls
some
illuminance
out
of
the
RGB
data
as
well.
You
have
a
I
had
a
lot
of
RGB
data,
so
why
not
use
that
the
luminance
inside
that
data
and
I'll
describe
how
that's
done
and
then,
after
that,
I
just
have
a
sequence
of
of
operations
to
generate
an
L
RGB.
You
know
a
color
image
and
running
over
that
quickly.
A
There's
you
know
you
put
the
RNG
and
B
together
I'm
going
to
do
this
in
detail,
but
just
to
say
it
in
30
seconds
put
that
stuff
together
the
usual
dbe
technique
for
getting
rid
of
gradients
and
finding
the
color.
You
know
setting
the
color
right,
the
the
Exterminator
plugins,
combining
the
luminance
with
the
RGB
and
then
operating.
A
You
know
if
the
image
most
of
my
images
do
if
the
image
calls
for
it
separating
out
the
main
target
from
the
stars
and
stretching
those
separately
and
then
combining
them
back
together
and
then
often
dealing
with
a
color
cast.
In
the
background,
most
of
my
images
don't
have
a
lot
of
you
know
clouds
shall
we
call
them
in
the
background
and
and
so
I'm
really
looking
or
or
if
they
do
I,
don't
pull
them
out
as
well
as
I.
Might
this
isn't
the
typical
narrow
band?
A
A
C
A
Okay,
so
that's
that's
probably
the
confusion
I
had
so
this
is
again
weighted
batch
process.
Pre-Processing
I'm,
just
mentioning
it
because
it
was
a
technique
that
was
valuable
to
me.
I,
don't
think
it
improved
in
a
big
way.
These
images,
but
as
long
as
I
was
giving
this
presentation
I
know
I
was
a
laggard
and
switching
over
to
it.
A
So
I'll
mention
this.
So
this
is
what
I
do.
This
is
a
more
complicated
version.
You
know
the
kind
of
image
for
wbp.
So
that's
why
I
picked
it
as
an
example
and
what
you
see
here
is
just
the
director
a
directory
structure,
a
bunch
of
file,
how
I
organize
files-
and
you
can
see
I-
have
a
directory
or
a
folder.
A
Whatever
word
you
like
to
use
called
P
for
pre-processing,
I,
guess,
M82
and
I
made
four
folders
inside
of
it
all,
starting
with
uppercase
session,
and
then
a
number
and
I
broke
up
the
data
that
I
shot.
You
know
after
I
cleared
it
through
blink
into
four
parts
and
I'll.
Explain
why
four
parts
so
these
two
sessions,
one
and
two-
were
the
two
two
sets
of
Subs
basically
related
to
two
sets
of
flats
that
I
took.
A
So
there
were
basically
I,
don't
know
three
weeks
or
four
weeks
that
I
shot.
This
object
and
I
only
take
a
set
of
flats
every
week
or
so
I,
don't
know
lazy.
So
there
you
go,
and
here
you
go.
December
13th
I
took
flats
and
these
images,
starting
on
December
14th
of
M82,
are
in
this
folder,
so
the
red,
green,
blue
and
luminance
of
M82
shot.
A
You
know
a
week
before
New
Year's
and
then
I
picked
it
up
again
on
January
19th
and
again
here's
my
January
19th
flats
and
my
image
data,
for
you
know,
probably
a
week
later.
For
that
week,
in
a
another
week
or
two
and
the
same
darks
as
before,
and
what
wbp
does
is
that
you
know
it,
it
will
process
these
Flats
with
these
darks
and
these
you
know
Subs
with
these
flats
and
darks
and
so
on
and
and
I
didn't
know
previous.
A
But
now
I
do
how
to
deal
with
multiple
sets
of
flats
and
so
on,
and
then
the
reason
I
have
a
session.
Three
and
four
is
just
because
the
AHA
data
is
taken
at
a
different
gain.
So
since
I
used
gain
139
for
the
narrow
band,
I,
didn't
you
know
that's
incompatible
with
this
75
dark,
so
I
just
separate
out
the
data
from
the
lrgb
data,
and
it's
again
this
is
December.
A
A
Okay,
what
I
do
and
I
may
be
a
lazy
one
is
I'll
shoot
flat.
I
mean
keep
in
mind
that
I
don't
move
my
telescope,
so
it's
in
the
back.
It's
covered
up
and
so
I'm
not
setting
up
my
telescope.
I
know:
Richard
goes
on
his
roof
and
and
resets
is
a
telescope
up
every
time
he
images,
so
he
may
want
to
take
Flats
every
time
he
images.
E
A
A
So
sorry
about
that
Richard.
In
any
event,
since
my
scope
is,
you
know
not
moving
around
a
lot
I'm
covering
it
and
uncovering
it
up,
I,
don't
feel
the
need
to
take
new
Flats
every
image
session,
so
I
will
take
flats
and
then
image
with
them
for
a
week
unless
something
changed.
You
know,
like
the
the
the
camera
rotated
relative
to
something
you
know
so
something
changed.
A
But
if
my
Optics
are
staying
the
same
I'm
not
retaking
my
flats
for
a
week
or
so
and
then
maybe
next
week,
I'll
take
them
again
and
I
typically
take
25
flats
for
each
filter
that
I'm
gonna
shoot
with
so
25
green
red,
blue
luminance.
In
this
case,
darks
I
shoot
like
once
a
year.
You
know
I
get
the
telescope
in
a
dark
place
and
I
know
you
know.
I
I
have
to
I
mean
it'll,
probably
be
a
night
when
there's
a
full
moon
out
and
I'll
just
shoot.
A
You
know
at
all
games
that
I,
typically
use
and
all
exposures
I,
typically
use
and
I
just
you
know,
and
then
I
run
through
the
processing
of
those
darks
and
I
saved
the
Masters
and
there
you
go
and
then
I
reuse,
that
for
the
next
year
or
so
anyway,
that's
what
I've
been
doing
I'm,
not
saying
I'm,
the
final
Authority
on
that.
But
and
then
you
know,
the
image
I
don't
know
depends
on
the
object.
A
How
many
Subs
take
you
know
something
like
a
cluster
where
you're,
basically
Imaging
Stars,
you
probably
don't
need
as
much
as
some
kind
of
dim
nebula.
You
know,
but
it
shows
you
on
my
slideshow.
What
I
did,
but
again
my
situation
is,
is
as
I'm
shooting
in
Suburban
Skies,
so
I
need
more
light
than
let's
say
if
you
were
shooting
at
Pinnacles.
You
know
probably
an
order
of
magnitude
more,
if
not
more
than
that
and
anyway,
if
that
answers
the
question,
that's
my
point
of
view.
A
All
right:
well,
if
nobody,
if
somebody
pipes
up
later,
that's
great
but
I'll,
go
on
since
I
haven't
heard
anything
so
I
was
going
to
show
a
little
bit
about
weight,
weighted,
batch
processing
and
then
move
on
to
this
image
here
so
weighted
batch
pre-processing
is
here
under
scripts.
Batch
processing
weighted
batch
pre-processing.
A
If
you
pull
it
up
like
I,
just
did
and
wait
here.
It
goes
I'm
going
to
reset
it
because
I
guess
I
left
it
working
and
you
can
see
here
well,
first
thing
I
would
do,
is
I
click
on
directory
down
here?
No,
let
me
tell
you
what
first
thing
I
do
is
I
make
sure
that
I
have
this
keyword
here
session.
You
saw
how
I
organize
my
data
by
these
directories
that
had
the
word
session
in
it.
A
So
you
want
to
make
sure
that
if
you
use
that
convention,
you
know,
of
course
the
word
itself
doesn't
matter.
You
just
have
to
be
consistent,
but
anyway,
so
you
can
see
here.
I
have
grouping
by
keyword
selected
with
session
and
I
showed
you
how
my
data
was
and
then
all
I'm
going
to
do
is
find
that
directory,
which
is
conveniently
already
here-
and
you
know
these
are
the
directories
that
I
showed
you
earlier
and
I
hit
open
and
weighted
batch
process.
Pre-Processing
just
ingests
all
that
data
you
could
see
it
says,
there's
1273
images.
A
It's
going
to
ingest,
that's
more
than
the
number
of
Subs
that
that
I
just
showed
you
because
of
all
the
flats.
So
it's
ingesting
the
subs
and
the
flats
and
the
master
darks,
the
darks.
It
isn't
gonna
build
because
they've
already
been
built,
but
there
you
go,
it
read
them
all.
If
I
hit
run,
it
will
take
10
hours,
I
mean
I
could
abort
it,
but
it
would
take
10
hours
to
run
if
you
hit
Diagnostics.
You
could
see
that
this
is
the
amount
of
data
it
expects.
A
It
expects
to
use
141,
gigabytes
or
140
gigabytes
of
scratch,
space
plus
the
existing
data
over
there.
But
anyway,
if
I
did
run
that
it
would
come
in
the
morning
or
whenever,
with
you
know,
the
images
I
could
need
to
start
doing
the
techniques
I'm
about
to
show
you
anyway,
that's
weighted
batch,
pre-processing
I'm
a
convert,
so
I
wanted
to
show
you
this
image.
You
may
recall
you
know
this
is
the
you
know
better.
Looking
M13
and
I'll
show
you
how
I
got
there
so
we'll
run
through
a
single
pre-processing.
B
A
On
the
second
image,
we'll
just
do
a
straight
process
of
this,
so
it
says
talking
right
over
the
linear
fidget
that
I
wanted
to
start
with.
So
the
first
thing
I
do
is:
is
linear
fit
my
channels
together,
so
we're
gonna
find
linear
fit
it's
funny.
How
that
that's
Murphy's
Law
so
I
know
that
my
green
channel
is
typically
the
strongest
channel,
so
I'm
going
to
linear
fit
the
other
channels
to
Green.
A
A
A
A
A
There
you
go
and-
and
we
have
that
so
at
this
point,
I
run
the
two.
Oh
no
at
this
point,
I
want
to
color
correct
the
RGB
image
and
once
you've
cropped
and
stuff
like
that,
you've
lost
your
what's
the
word,
you
know
the
ra
deck
position
rotation,
and
so
you
have
to
solve
the
image
again
and
under
scripts
image.
Analysis
you'll
find
image
solver
and
you
can
run
that
and
solve
your
image.
A
B
A
A
D
A
A
A
A
Thank
you
and
but
I
will
say
that
the
one
thing
that
can
sometimes
hit
me
here
is,
if
my
image,
for
whatever
reason,
is
lower
quality
than
some.
You
won't
see
this
nice
line.
You
know
you'll,
see
much
more
of
a
scatter
or
you'll,
see
very
few
points
and
if
that's
the
case,
I
often
push
this
later
on
in
the
processing
to
you
know
after
the
noise
removal
or
such
or
you
know
anyway,
but
in
this
case
it
looked
fine
and
now
we'll
go.
A
Now
the
next
step
in
this
that's
done.
The
next
step
in
this
is
to
combine
the
luminance
and
RGB
data
and
using
this
lrgb
combination
process
and
picks
Insight.
Now,
that's
supposedly
you're
supposed
to
do
that
in
on
non-linear
images,
but
I
wanna
I
want
to
wind
up
with
a
a
linear
image,
because
I'm
going
to
do
my
stretching
after
I
do
that
so
you
still
unmuted
Francesco.
So
what
I
do
as
an
inverted
an
invertible
stretch.
A
So
if
you
look
at
this
histogram
stretch
the
shadows
and
the
and
the
highlights
are
you
know
zero
and
one,
but
the
mid
tones
is
a
0.02
which
you
can
invert
it's
just
a
curve
and
you
can
run
the
inverse
curve
on
it.
So
that's
what
I
do
I
I
stretch
both
these
guys
so
that
they
become
non-linear
and
then
I
run.
Lrgb
combination
with
the
synthetic
just
just
I'm
just
gonna
add
the
synthetic
image
back
into
the
art.
A
A
And
there
you
go
voila
and
and
that's
the
stf
and
that's
what
a
typical
you
know.
Image
will
look
like
all
of
a
sudden
with
all
this
stretching
and
noise
reduction
and
whatnot
you
wind
up
seeing
I
mean
the
stf
is
really
bringing
out
the
noise
in
the
background,
but
unfortunately
that
stuff
tends
to
go
away,
but
anyway,.
B
A
A
All
I'm
going
to
do
here
is
is
stretch
and,
and
we'll
see
so,
let's
stretch
it
so
first
thing
I'm
going
to
do
is
get
rid
of
the
auto
stretch,
because
this
ghs
doesn't
want
to
see
that
and
we're
going
to
use
what
David
Payne
taught
us
last
month
about
how
to
remove
sort
of
at
least
part
of
the
offset
that
the
image
has.
So
if,
if
you
look
here,
you
could
see
that
there's
a
lot
of
nothing
and
then
there
an
image
starts.
A
Payne
taught
us
that
if
you
click
here
hit
send
to
SP
set
the
protect
highlights
down
to
zero
I
mean
to
as
low
as
it
goes
anyway
and
then
move
this
back
all
the
way
to
the
left.
If
you
increase
the
stretch
factor
at
that
point,
you
can
sort
of
move
the
image
all
the
way
to
the
left.
So
I
do
that,
just
not
to
waste
any
pixels,
I
guess
and
that's
I,
don't
know
if
it's
necessary.
It
seems
like
a
good
idea
and
now
it
starts
the
rail
stretching.
A
So
to
start
the
real
stretching
you
click
the
real
time
preview.
You
put
your
symmetry
point.
You
know
near
this
peak
here,
click
Send
to
SP
and
then
play
with
the
stretch,
factor
local
intensity
and
symmetry
Point.
Fundamentally
is
the
scheme
that
I've
grown
to
love
and
so
I.
Don't
know
you
wind
up
with
your
favorite
numbers.
My
favorite
numbers
for
images
like
this
tend
to
be
in
this
kind
of
ballpark.
Here
we'll
pick
something
like
that.
David
showed
us
how
he
looks
at
the
curve
here
and
adjust
it.
A
I've
honestly
have
been
just
looking
at
the
picture.
So
let's
do
that.
Here's
a
what
seems
like
a
reasonable
start,
oh
by
the
way,
just
qualitatively
this
local.
You
know
qualitatively
the
stretch
factor
right.
You
know
the
more
to
the
right.
You
go,
the
brighter
things
go,
of
course,
and
the
local
intensity
seems
to
be
like
in
a
flatter
image.
If
you're
to
the
right,
whereas
you
know
you
move
it
more
to
the
left
at
least
till
you
get
to
zero
the
brighter
things
get
even
brighter.
A
That's,
what
I
would
say
qualitatively
when
I
wind
up
in
Middle
grounds
like
this,
let's
say
and
and
then
what
I
do
is
I
play
with
the
I
adjust
the
SP
symmetry
point
so
that
you
know
if
you
bring
it
to
the
left,
you're
getting
more
background.
If
you
bring
it
to
the
right,
you
know
your
background
is
getting
darker,
which
is
good,
but
you
don't
want
your
background
so
dark
that
you
start
losing
detail
in
your
image.
You
know,
like
hiding
parts
that
you
really
want
to
see.
A
So
that's
sort
of
an
eye
thing
and
I'm
just
gonna
go
around
here,
and
this
image
is
close
to
done
just
by
doing
that,
so
the
only
there
you
go
and
I'm
just
gonna,
that's
all
the
stretching
I'm
gonna
do
in
that
I'm
going
to
get
rid
of
the
real
time
preview
and
here's
my
stretched
star
cluster.
Honestly,
it's
it's
close
to
a
final
image
for
me.
Maybe
I
would
do
some
Lightroom
adjustments
or
something
like
that.
A
The
only
other
thing
that
I
tend
to
do
and
I'm
going
to
demonstrate
it
right
here
is
desaturate
the
background.
So
I
don't
know
if
it
is
a
byproduct
of
you
know
a
poor
Imaging
or
what
but
I
tend
to
get
like
this
image
has
red.
You
know,
in
the
background,
the
background
is
red
a
little
bit
but
noticeable
to
me
and
noticeable
enough
that
I
want
to
do
something
about
it.
A
So
what
I
do
is
honestly
just
use
the
process
which
is
desaturate
along
with
a
mask
that
masks
out
all
the
bright
things
in
the
image
so
I'm
going
to
do
that,
just
to
demonstrate
how
I
do
that,
so
to
do
that,
I.
Take
my
here's,
my
image
I'm
going
to
extract
the
luminance
from
it,
and
so
here
it
is
I'm.
A
So
this
is
my
mask.
There
is
my
RGB
data
I'm
just
going
to
take
my
RGB
image
under
mask,
select,
mask
and
select
that
and
invert
that
mask
and
there
it
is,
and
then
one
sec
I
find
I
have
the
saturation
tool
here.
I
raise
the
range
to
10,
believe
it
or
not,
and
I'm
gonna
do
this
to
about
minus
eight.
If
I
find
it
and
apply
it
there,
you
go
done
and
that's
it
for
this
image,
I'll
get
rid
of
the
Mask.
A
A
Let's
see
here
if
I
go
to
histogram,
transform
and
I.
Look
at
this
image.
You'll
see
that
there's
a
lot
of
nothing
at
the
start.
So
I
can
move
this.
You
know
keeping
an
eye
over
here
on
how
many
pixels,
I'm
clipping
and
I
can
move
the
you
know.
There's
you
know
20
million
pixels
or
whatever
in
this
image.
So
if
I
flip
a
thousand
of
them,
that's
really
nothing.
A
A
A
C
So
go
ahead:
fire
go
ahead,
so
I
know
that
in
general
people
that
I've
spoken
to
have
said
that
for
generating
a
flat,
they
typically
use
a
an
ordinary,
inexpensive,
Chinese,
let's
say
slight
illumination
panel
and
at
first
that
made
sense
to
me.
But
then
I
saw
ads
on
beta
planetarium
that
they
have
a
a
flat
tool
that
is
actually
an
electroluminescent
panel,
and
the
difference
is
that
a
regular
ordinary
panel,
an
inexpensive
panel,
even
though
it
appears
more
or
less
white
to
the
eye.
It's
really
very
few
spectral
lines.
C
It's
not
a
flat
of
a
generation
of
light
over
many
colors.
It's
it's
just
a
combination
of
a
couple
of
spectral
lines
and
that
in
fact,
is
also
what's
going
on
in
fluorescent
light
bulbs
is
that
they
only
generate
a
few
lines
and
then
and
then
your
eye
thinks
that
it's
white
well,
the
beta
is
claiming
that
they
could
do
a
better
job
by
using
this
electroluminescent
panel.
That's
generating
true
wide
spectrum
flat
white
light.
D
Yeah,
that
seems
to
work
for
me
too.
I
had
problems
when
I
was
using
a
a
tablet
for
Flats,
because
in
that
case,
the
the
colors,
the
white
from
the
background
LEDs
are
filtered
via
B3,
rgmb,
pixels,
and
so
what
you
get
is
the
combination
of
the
Spectra
and
was
giving
me
some
strange
problems
with
the
camera.
But
you
know
with
the
Chinese
panel:
I
didn't
have
any
problem.
D
Actually
I
I
did
what
something
similar
to
what
Julian
told
us
in
one
of
the
previous
presentation
and
I
put
an
Arduino
to
control
the
brightness,
and
it
works
pretty
well
the
El
panel
I,
the
ones
that
I
read
about.
They
are
kind
of
a
weak
in
illumination,
so
they
require
a
longer
very
long.
Exposure
for
for
narrow,
Bend
I,
don't
know
if
Baylor
has
something
better.
There.
C
I
didn't
check
on
the
actual
brightness
value
for
the
better
panels,
but
it
just
from
a
physics
point
of
view.
It
seemed
to
make
sense
to
me
that
if
you're
only
pretending
that
you're
generating
white
light,
but
in
fact
you're
generating
narrow
spectral
lines
and
only
a
few
of
them
that
simply
tricked
the
brain
by
adding
together
to
produce
white
in
our
brain,
then
I
thought.
Well,
then
it's
not
really
calibrating
flat
for
all
colors
that
could
be
presented
at
the
objective
of
the
telescope.
Yeah.
D
F
C
C
The
other
question
I
had
was
for
specifically
for
high
I'm,
wondering
I
noticed
that
the
images
varied
from
one
to
another
and
particularly
I
noticed
the
refractor
image
that
you
presented
had
extremely
tight,
sharp
round
stars
all
over
the
field,
and
the
reflector
would
vary
from
one
image
to
another
and
also
from
one
side
of
the
image
to
the
other
and
I'm
wondering
if
that
could
be
a
collimation
issue
or
whether
it
could
be
a
a
back
focus
issue
or
it
could
be
a
field.
Flattener
issue,
yeah.
A
C
A
Don't
have
a
fill
flattener
on
it
so
and
I
did
have
a
field
flattener
on
the
refractor
and
the
refractor
I'm
sure
is
collimated
better
than
the
RC
I
mean
RC
is
a
tricky
thing
to
collimate,
so
I
don't
tend
to
be
a
pixel
people,
I
mean
I'm.
Not.
This
is
not
a
knock
on
looking
at
the
tight
corners
and
all
that
I
have
no
doubt
that
I.
F
A
Terms
of
tight
Stars,
you
know
the
seeing
is
you
know
the
the
the
RC
is
0.3
Arc
seconds
per
pixel,
which
is
ridiculous
because
my
seeing
is
probably
two
Arc
seconds
and
or
three
you
know
and
the
refractors
closer
to
you
know
reality.
You
know
like
it's
probably
a
little
over
one
or
a
second
per
pixel.
A
All
right
and
and
then
you
know
there,
you
have,
you
know
blur
X
trying
to
improve
these
things
right.
You
know,
so
it's
taking
the
you
know
the
whatever
distorted,
RC
I
mean
they're
all
distorted,
but
let's
say
the
more
distorted
RC
and
then
it's
trying
to
fix
that
data
up
and
it's
neural
net,
which
is
an
interesting
thing.
A
So
anyway,
let
me
move
on
I
was
you
know
anyway,
and
show
the
the
how
how
I
created
this
synthetic
luminance
the
scheme
that
we
came
up
with
a
while
back
the
idea
being
look
I
have
this
rgmb
data
and
it
has
luminance
in
it
and
then
I
shot.
You
know,
50
of
my
Imaging
time
was
that
a
little
less
you
know
less
than
50
percent
of
the
photons
I
would
have
thought.
Is
this
RGB?
Because
these
channels
are,
you
know,
have
smaller
bandwidth
and
the
luminance?
A
But
anyway
there
is
significant
number
of
photons
coming
in
there.
You'd
think
that
I
should
be
able
to
get
detail
from
them
as
well,
not
just
throw
them
away
in
the
lrgb
combination.
So
the
idea
was
to
combine
the
luminance
from
the
RGB
with
the
luminance
that
I
shot.
That
was
the
hope,
and
so
here's
the
way
I
did
that
so
like
before
I'm
going
to
to
the
linear
fit
should
be
set
up
for
green,
so
like
before
I
linear,
fit
the
red
and
blue
channels.
A
To
the
green
so
that
they're
at
the
same
level
and
light
like
before
I'm,
going
to
combine
the
the
channels
together
and
I
have
this
RGB
image
over
here:
let's
I'm
having
trouble
only
because
Zoom
is
interfering
with
me,
but
okay
did
it
and
then
there's
this
you
can
see
it's
it's
anyway
there.
It
is
what
it
is.
There's
this
recommendation,
I
forget
I,
read
where,
where,
if
you're
going
to
do
something
like
I'm
about
to,
do
you
make
sure
each
Channel
contributes
equally
so
I?
A
For
now,
let's
move
on
and
now
I
have
this
RGB
image
and
I'm
going
to
extract
the
L
from
it,
and
so
now
I
have
two
L
channels.
I
have
this
l,
which
is
from
the
RGB
data
and
I,
have
my
friend
the
L
that
we,
you
know
pre-processed
and
and
shot
the
original,
L
Channel
and
so
more
or
less.
What
I
want
to
do
is
average
these
two
channels,
or
at
least
this
is
what
I've
been
doing
and
how
so,
how
do
you
average
channels?
Well,
image
integration
effectively
averages
channels.
A
A
overwriting
the
last
time
I
did
this
okie
dokie
and
I'm
gonna
do
the
same
thing
for
my
L
Channel,
which
is
already
on
disk,
but
I,
don't
remember
where
so
I'm
going
to
save
him,
yes
to
Temp,
L
sure
enough!
It's,
Already
Done
must
have
practiced
this
talk
and
there
you
go
so
now.
I
have
them:
okay,
whatever
so
now,
I
have
them
saved
to
disk,
and
now
I
can
use
the
add
files
button
and
there's
temp
L
and
temp
lrgb
and
I
read
them
both
in.
A
A
Okay,
so
here's
my
image,
integration
and
I,
rename
him
Elson
and
that's
the
synthetic
L
that
we're
going
to
use
for
the
processing.
So
that's
the
technique
to
create
my
synthetic
L
again,
it's
you
know
pretty
similar
to
my
L
channel.
It's
just
that
I've
also
added
into
it.
The
luminance
data
sort
of
derived
from
my
color,
my
color
channels
as
well.
I,
don't
need
this
guy
anymore.
A
A
And
now
we're
just
going
to
do
the
same
old
processing.
We
did
that
I
showed
you
before,
hopefully
quickly
and
I.
Think
we'll
probably
call
this
a
day
after
that
so
I'll
solve
and
this
time
I
don't
have
to
bring
it
out.
You
know
from
the
menus,
I'll
just
solve
it
by
dragging
it
and
then
I'll
get
the
colors
right
over
here.
A
There
he
is,
the
color
is
significantly
better
I'm,
going
to
run
dbe
on
both
and
I've
pre-done
that
if
anybody
needs
a
lesson,
you
know
you
can
ask
later
on
how
to
get
these
samples.
I
I
took
Rob
file's
suggestion
of
using
a
lot
of
these
samples,
at
least
when
you're,
not
interested
in
the
you
know,
like
I,
say:
finding
lots
of
detail
in
background
clouds,
so
I
used
40
samples
across
and
up
and
down.
A
B
A
A
When
I,
you
know
when
it's
not
a
practice
process
like
this
one
is
tonight,
I
would
say
running
through
all
these
steps
will
take
me
a
couple
hours.
You'd
think
it
would
take
half
hour,
but
there's
always
some
complication.
You
know
that
gets
you
to
think
a
little
bit,
but
it
it's
not
too
bad.
Okay.
A
A
A
Okay,
so
here
you
go,
this
is
the
RGB
Stars
image.
What
I
call
RGB
Stars,
which
is
you
know,
obviously
just
the
Stars
extracted-
and
this
is
the
Galaxy
I'm
gonna-
make
it
a
little
bigger
and
just
point
out.
It
did
a
good
job.
It
typically
does
a
good
job
in
my
experience
slightly
better
than
starnet
plus
plus,
but
you
could
use
that
too.
A
You
can
see
over
here
remnants
of
a
star.
You
know
sort
of
the
brightness
Parts
I
I
used
to
pull
it
out
of
Photoshop
pull
this
image
out,
and
you
know
correct
this
part
of
the
background.
Remove
you
know
heal.
If
you
will
this
star
Remnant
out
I
found
that
it
doesn't
make
that
much
of
a
difference.
You
can
probably
see
it,
but,
and
then
anyway,
it
sort
of
worked
out
of
my
workflow
I
could
show
you
the
difference.
If
you
do
that
or
don't,
but.
B
A
There
you
go
so
here
is
the
Galaxy
and
there
are
the
stars.
Auto
stretched,
but
we're
gonna
do
a
better
job
stretching
them,
and
then
the
next
step
is
to
stretch
so
I'm,
going
to
remove
the
screen,
transfer
function
and
then
get
to
the
stretching
and
so
here
again
I'm
gonna.
Do
that
same
trick
that
Dave
Payne
taught
us
about
removing
the
oh
there.
G
E
E
A
Well:
okay:
there
you
go
no
Darkness
Noise
Okay
anyway.
I've
done
that
and
it's
religion
for
me.
So
here
you
go,
I
I
think
I
hit,
hopefully
I
believe
I
did
right,
I
hit.
Well.
Let
me
check
I'll
check
the
history,
Explorer
I
sort
of
lost
my
contacts
RGB.
Yes,
so
I've
already
done
one
stretch
and
that
stretch
was
just
this
silly
little
thing
that
we
just
we're
talking
about.
So
now
we
move
to
the
real
stretching
so
I
remember
to
hit
reset
I
zoom
in
a
little
bit.
A
A
A
You
know
looking
at
the
you
know
like
when
I
set
like
here,
you
could
see
I
setting
the
S
point
a
little
too
high
here
right
and
we're
losing
some
of
the
glow
around
the
Galaxy
and
if
I
move
it
to
the
left,
you
know
the
background
is
is
lighter
than
I
would
like.
So
that's
sort
of
the
thing
I'm
looking
at
you
know
I
yeah,
something
like
this
looks.
Okay,
honestly,
that's
the
sort
of
fudge
Factor
there
and
then
these
numbers
I,
don't
know
that
looks
a
little
better
to
me.
A
Anyway.
It's
it's
playing
around.
It's
probably
not
not
mattering
that
much.
You
could
probably
see
that
that
thing
that
I
was
talking
about
did
come
out
a
little
bit
that
that
little
Galaxy
came
out
nicely
anyway.
That's
that
for
that
now,
once
I
do
that
once
I
have
an
okay
first
stretch,
what
I
tend
to
do
is
use
the
two
sort
of
you
know:
artificial
contrast,
things
HDR,
Mt
and
local
histogram.
A
I
am
a
little
troubled
by
the
fact
that
my
RGB
is
black,
so
maybe
I
forgot
to
apply
that
dress
yeah,
so
I
forgot
to
apply
that
stretch.
So
let
me
do
that
again.
I'm,
sorry,
so
I
set
it
all
up
and
then
I
didn't
apply.
It
that's
a
mistake.
So
here
you
go.
This
guy
set
to
M10
I
had
him
around
seven
and
I'm
around
two
or
three.
B
A
You
know
you
just
don't
want
to
start.
You
know
once
you
see
that
Halo
shrinking
you're
too
far
to
the
right
on
the
SP
parameter,
but
you
know
you
do
want
to
get
the
I
mean.
That's
the
reason
that
these
stretches
are
better
because
you
know
we're
not
stretching
that
part
to
the
left,
where
all
the
noise
is
you're.
Only
you
know
you're
stretching
the
parts
after
it
a
lot
more.
A
That's
the
advantage
of
the
general
hyperbolic
stretch,
I,
suppose
and
I
guess
I
liked
it
a
little
better,
but
this
is
a
little
lower
okay.
So
this
is
I
believe
where
about
where
I
was,
and
this
time
I
remember
to
actually
stretch
it
and
get
out
of
here
and
then
here
we
go.
There's
the
RGB
image
and,
as
I
said,
I
tend
to
like
these
two.
A
A
A
Okay,
so
there's
our
Galaxy
and
I
would
say
that
you
know
roughly
I'm
done
with
the
Galaxy.
Some
folks,
maybe
could
could
do
more
with
that,
but
I'm
going
to
call
that
a
day
and
start
working
on
the
Stars.
So
you
may
recall
that
I
stretched
I
mean
when
I
did
this
I
separated
the
stars
and
I
had
there
should
be
an
image
somewhere,
I
don't
see
called
RGB
stars
there.
It
is,
doesn't
look
quite
this
ugly.
A
It
is
a
little
green,
interestingly,
which
I'm
going
to
get
rid
of
later
so
I'm
going
to
undo
the
stretch
and
go
into
the
general
hyperbolic
stretch
again
and
for
I.
Remember
Francesco.
In
his
talk
said
he
used
six
and
two.
You
know
I
think
what
what
happens
is
we
find
these
numbers?
We
like
and
I,
don't
know
Francesco
correct
me:
I,
don't
want
to
put
words
in
your
mouth:
okay,
I,
wind
up
with
I,
don't
know,
maybe
a
little
stronger
stretches.
I
don't
know!
Maybe
I
I
have
this.
A
What
I
wind
up
doing
yeah
you
know
like
the
two
is,
is
not
a
bad
Target
for
me
for
that.
But
what
I
wind
up
doing
is
pulling
this
up
I
like
to
see
a
lot
of
stars.
You
know
so
like,
for
instance,
this
I
think
is
under
stretched.
There's
a
lot
of
stars.
You
don't
see
at
all
at
least
I,
don't
on
the
screen,
and
now
here
I
start
seeing
all
these
dim
stars
and
then,
if
I
do
too
much,
you
know
then
I
get
all
these
distorted.
A
You
know,
especially
probably
because
of
my
ASI
1600
camera,
distorted
thing
so
I'm.
You
know
I
try
to
find
this
region
between.
You
know
you're,
seeing
the
dim
stars
and
the
real
you
know,
brighter
ones
aren't
too
distorted.
So
let's
say
something
like
that.
So
we'll
just
call
that
a
day
I
there
is
not
none
of
this.
You
know
getting
rid
of
the
the
part
on
the
left
for
the
stars,
because
I
think
the
star
extraction
has
taken
care
of
that
and
for
us
hi.
A
Oh
yes,
I
I
forgot,
you
know
in
fact,
for
both
I
usually
use
color
and
so
I
made
a
mistake
there.
So
let's
do
that
I'm
gonna!
That's
a
good
point!
Thank!
You
just
rush
through
that.
So
you
know
this
has
too
much
color,
probably
setting
the
color
mode
to
a
hundred
percent,
but
I
don't
know.
Let's
say:
25
looks
okay
to
me
on
this
image.
A
It's
bugging
me
just
by
the
scnr
program
and
I'm,
going
to
call
those
Stars
stretched
I'm
going
to
show
you
what
I
learned
was
on
this.
How
to
you
know
realize
oh
I
forgot
something
that
I
really
wanted
to
do
without
spending
all
that
time
and
re-stretching
this
a
third
time
because
I'm
always
forgetting
something
I
really
wanted
to
do
so.
I
go
to
the
history
Explorer
and
first
of
all,
you
can
pull
out
from
the
history
Explorer
this
and
keep
alive
and
and
find
all
the
hist.
A
You
know
find
what
you've
done
and
I'm
just
gonna
set
him
aside
and
then
I'm
gonna
go
to
the
history
Explorer
and
go
back
to
the
point
where
I've
done
that
you
know
that
first
stretch
just
to
get
rid
of
the
stuff
on
the
left
and
I
set
up
and
then
I'm,
gonna,
double
or
I
can
do
it
here.
Double
click
on
this
guy
and
now
he
is
stretching
the
we're
back
to
you
know
before
that
main
stretch
where
I
forgot,
to
put
it
in
color
and
I.
A
Do
that
and
yes,
you
can
see
it's
a
lot
more
colorful
again
too
colorful,
okay,
but
more
colorful
and
we'll
zoom
in
and
okay,
maybe
that's
too
much
red,
but
it
would
be
nice
to
have
some
of
it.
And
so
you
know
something
like
that.
Maybe
in
this
case,
half
I'm,
okay
with
so
thank
you
Francesco
for
that
and
that
that's
how
I
go
back
and
now,
of
course,
I'm
just
gonna
drag
in
the
same
hdrmt
and
the
same
local
histogram.
That
I
did
before.
A
And
here
I
am
I'm
done
with
the
stretch
of
the
Galaxy
I'm
done
with
stretching
of
the
stars
and
I
use
the
pixel
math
equation.
I
learned
from
Francesco
to
combine
them,
which
is
this
one
here
dollar
t
means
the
image
itself.
So
I'm
gonna
drag
this
pixel
math
over
this
image.
And
basically
you
know
not
not
this
image
times,
not
the
Stars
winds
up
screening
in
the
stars,
I
guess
so
here
we
go
and
that's
pretty
close
to
my
final
image.
A
I
think
I
still
have
you
know:
I
I
still
have
empty
space
on
the
left
and
I
bet
we'll
take
a
look
and
sure
enough.
You
know
you
don't
need
this.
This
is
the
kind
of
thing
I
could
also
get
rid
of
in
a
Lightroom,
but
why
not
get
rid
of
it
here?
You
can
actually
see
here
how
many
pixels
you're
clipping.
A
So
you
know
some
number
like
that
is
I
believe
not
too
harmful
to
the
image
and
there
you
go
and
there's
my
you
know
much
better
than
last
year's
version
of
NGC
891,
so
I've
been
going
on
for
90
minutes,
so
I
could
do
another
m33,
but
the
it's
a
very
similar
technique.
The
only
thing
I
was
going
to
add
with
m33
is
that
it
was
an
aha
image
and
what
I
did
was
I
I
guess
I
could
skip
ahead
of
people
one
and
but
I'm
happy
to
end
here.