Tutorial on Stacking and H-alpha Imaging

Images at
http://home.att.net/~dpersyk/new.htm

I have prepared a series of images of IC 468 (Thor's Helmet) that
illustrate the benefits of stacking images. Images with total
integrations from 10 seconds to 4200 seconds are presented. Scope is
4-inch refractor, camera MX716.

I also show images with and without an H-alpha filter. The contrasts
are striking. I hope you will take a look and find the page
informative.

Clear skies,

Dennis Persyk
Igloo Observatory Home Page http://dpersyk.home.att.net
Hampshire, IL

(Dennis Persyk) wrote:

Images at
http://home.att.net/~dpersyk/new.htm

I have prepared a series of images of IC 468 (Thor's Helmet) that
illustrate the benefits of stacking images. Images with total
integrations from 10 seconds to 4200 seconds are presented. Scope is
4-inch refractor, camera MX716.

I also show images with and without an H-alpha filter. The contrasts
are striking. I hope you will take a look and find the page
informative.

Most interesting, not to mention reassuring for those of us without
double-digit apertures who are interested in imaging DSOs! Having seen
you results a H-alpha filter just moved well up my wish-list. Thanks
for putting it out there for us all to learn from and be inspired by.

I'm curious about the (apparent) background noise on the 10s H-alpha
image - is that a result of the H-alpha filter dimming the image and
the noise being emphasised by you cranking up the brightness in
post-processing to compensate?


Tim
--
Love is a travelator.

Tim Auton tim.auton@uton.[groupSexWithoutTheY] wrote in message . ..
(Dennis Persyk) wrote:

Images at
http://home.att.net/~dpersyk/new.htm

I have prepared a series of images of IC 468 (Thor's Helmet) that
illustrate the benefits of stacking images. Images with total
integrations from 10 seconds to 4200 seconds are presented. Scope is
4-inch refractor, camera MX716.

I also show images with and without an H-alpha filter. The contrasts
are striking. I hope you will take a look and find the page
informative.

Most interesting, not to mention reassuring for those of us without
double-digit apertures who are interested in imaging DSOs! Having seen
you results a H-alpha filter just moved well up my wish-list. Thanks
for putting it out there for us all to learn from and be inspired by.

I'm curious about the (apparent) background noise on the 10s H-alpha
image - is that a result of the H-alpha filter dimming the image and
the noise being emphasised by you cranking up the brightness in
post-processing to compensate?


Tim

Hi Tim,

Thanks for checking out my web page. The H-alpha filter is truly the
world's best light pollution filter if one is imaging emission
nebulae.

The noise grains are individual dark (thermally emitted) electrons. I
did do a dark frame subtract, but, of course, the distribution of dark
electrons is purely random. Thus from one 10 second image to the next,
their distribution will vary spatially. Hence for short exposures
with a low dark current camera like the MX716, dark frame subtraction
does not really help. And yes, I have performed very, very aggressive
histogram stretching on all of the images.

Clear skies,

Dennis

On 21 Mar 2004 09:36:15 -0800, (Dennis Persyk) wrote:

The noise grains are individual dark (thermally emitted) electrons. I
did do a dark frame subtract, but, of course, the distribution of dark
electrons is purely random. Thus from one 10 second image to the next,
their distribution will vary spatially. Hence for short exposures
with a low dark current camera like the MX716, dark frame subtraction
does not really help. And yes, I have performed very, very aggressive
histogram stretching on all of the images.

Some clarification might be in order. Dark current is constant at any given
temperature, and dark current noise increases as the square root of that. The
dark current signal is perfectly subtracted, leaving only the dark current noise
behind. The signal accumulates faster than the dark current noise, so the S/N
increases with exposure time. In terms of thermal noise, there is no difference
between a single long exposure and a stack of very short ones of equivalent
time.

The noise you are seeing in the short exposures is not thermal noise, but
readout noise. Readout noise is what ultimately determines the minimum possible
exposure time, since it is a constant independent of time. With a high thermal
noise camera, or under light skies, readout noise is quickly swamped by other
noise sources, and there is no disadvantage to short exposures. With a well
cooled camera under dark skies, however, individual exposure times of 30 minutes
or longer may be necessary before the effects of readout noise are made
insignificant.

_________________________________________________

Chris L Peterson
Cloudbait Observatory
http://www.cloudbait.com

Chris L Peterson wrote in message . ..
On 21 Mar 2004 09:36:15 -0800, (Dennis Persyk) wrote:

The noise grains are individual dark (thermally emitted) electrons. I
did do a dark frame subtract, but, of course, the distribution of dark
electrons is purely random. Thus from one 10 second image to the next,
their distribution will vary spatially. Hence for short exposures
with a low dark current camera like the MX716, dark frame subtraction
does not really help. And yes, I have performed very, very aggressive
histogram stretching on all of the images.

Some clarification might be in order. Dark current is constant at any given
temperature, and dark current noise increases as the square root of that. The
dark current signal is perfectly subtracted, leaving only the dark current noise
behind. The signal accumulates faster than the dark current noise, so the S/N
increases with exposure time. In terms of thermal noise, there is no difference
between a single long exposure and a stack of very short ones of equivalent
time.

The noise you are seeing in the short exposures is not thermal noise, but
readout noise. Readout noise is what ultimately determines the minimum possible
exposure time, since it is a constant independent of time. With a high thermal
noise camera, or under light skies, readout noise is quickly swamped by other
noise sources, and there is no disadvantage to short exposures. With a well
cooled camera under dark skies, however, individual exposure times of 30 minutes
or longer may be necessary before the effects of readout noise are made
insignificant.

_________________________________________________

Chris L Peterson
Cloudbait Observatory
http://www.cloudbait.com

Hi Chris,

After looking at the statistics in dark frames of 0.1, 1, 10 and 100
seconds, I see that you are correct. They all have nearly the same
noise properties, so indeed at low signal levels the readout noise
predominates. Thank you for the correct interpretation.

Clear skies,

Dennis

Tim Auton tim.auton@uton.[groupSexWithoutTheY] writes:

I'm curious about the (apparent) background noise on the 10s H-alpha
image - is that a result of the H-alpha filter dimming the image and
the noise being emphasised by you cranking up the brightness in
post-processing to compensate?

Surely the difference in noise -as well as in signal- between the
unfiletered and Ha-filtered 10s images is different processing of the
two images? The Ha-filtered image must be "pressed" harder than the
unfiltered one to bring out details. After all, a filter cannot
increase the signal, as the comparison would seem to suggest.

pej
--
Per Erik Jorde

Per Erik Jorde wrote in message ...
Tim Auton tim.auton@uton.[groupSexWithoutTheY] writes:

I'm curious about the (apparent) background noise on the 10s H-alpha
image - is that a result of the H-alpha filter dimming the image and
the noise being emphasised by you cranking up the brightness in
post-processing to compensate?

Surely the difference in noise -as well as in signal- between the
unfiletered and Ha-filtered 10s images is different processing of the
two images? The Ha-filtered image must be "pressed" harder than the
unfiltered one to bring out details. After all, a filter cannot
increase the signal, as the comparison would seem to suggest.

pej

The processing is indeed different between the images, but Chris'
point about readout noise is very important. Let's consider the 10
second no filter and 10 second H-a filter images.
http://home.att.net/~dpersyk/new.htm

Bear in mind that the images were acquired under a full moon, with a
visual limiting magnitude of 1.5.

In the case of the filterless image, I processed by moving the black
point to where the sky was dark. No matter how I processed, I could
not bring out any of the luminosity of the nebula because it was
swamped out by moonlight. The target-to-background ratio was very
poor. The moonlight also swamped out the readout noise, so we do not
see any of the graininess.

Now, in the H-a filtered image, the filter is passing virtually only
light from the emission nebula, and the moonlight and other sky
background is nearly zero. I processed so as to show that some
nebulosity was now visible, and in so doing, I accentuated the read
noise. The nebula's emission signal level in the filtered image is
virtually the same as in the unfiltered image but the _background_ is
orders of magnitude less in the filtered image. That is the power of
hydrogen-alpha imaging of emission nebulae.

Thanks for an interesting discussion. I have learned a lot.

Clear skies,

Dennis
Igloo Observatory Home Page http://dpersyk.home.att.net
Hampshire, IL

Tim Auton tim.auton@uton.[groupSexWithoutTheY] wrote:
(Dennis Persyk) wrote:
Images at
http://home.att.net/~dpersyk/new.htm
[snip]
I'm curious about the (apparent) background noise on the 10s H-alpha
image - is that a result of the H-alpha filter dimming the image and
the noise being emphasised by you cranking up the brightness in
post-processing to compensate?

[loads of good answers and discussion]

Well, I'll take that all as pretty much a yes then

But I now know exactly what kind of a yes and why, and how I would
better phrase the same question next time (not that I'll need to ask
again). I probably should have been clearer that I meant background
noise relative to the unfiltered 10s exposure, but everyone seemed to
understand what I was getting at anyway.

Most informative, thanks all.


Tim
--
Love is a travelator.

On 22 Mar 2004 09:58:02 +0100, Per Erik Jorde wrote:

Surely the difference in noise -as well as in signal- between the
unfiletered and Ha-filtered 10s images is different processing of the
two images? The Ha-filtered image must be "pressed" harder than the
unfiltered one to bring out details. After all, a filter cannot
increase the signal, as the comparison would seem to suggest.

For short exposures, the Ha filter does not provide much advantage, since the
image is dominated by readout noise. For long exposures, however, where
(assuming a low thermal noise camera) the noise is dominated by sky background,
the Ha filter provides a better S/N (even though, as you correctly note, the
total signal is reduced).

_________________________________________________

Chris L Peterson
Cloudbait Observatory
http://www.cloudbait.com