| Long Exposure DSO Photo Techniques, Page 2 |
Capturing the Light
Two stages encompass the image creation, photon collection, and post collection
processing. There are a few key items that result in good photon collection, select the
appropriate FOV (the right scope/lens FL for your camera), frame your subject as per the
photographic rule of thirds (there are exceptions), expose long enough to bring out the
background interests and swamp read noise (but not so long as to raise the sky
background too high or compress/saturate the scene highlights), select a sufficiently bright
(high s/n) guide star for reasonable guiding, properly set your mount guiding parameters,
and collect enough frames through each filter to yield a smooth field and allow stretching.
As for photographic rules, some apply, some don't. Remember that you are taking a picture
of some, specific subject, not just a chunk of the sky. Declutter...bring your subject to the
image foreground, push the clutter into the background. But everything is at the same
distance photographically, you say. Highlight your primary subject, de-emphasize the
unnecessary clutter, you know what your subject is, and what is cluttering up the photo.
Collect lots of photons through each of the filters that you are using (lum, chrome,
narrowband; 3 to 4 hrs of lum, 1 1/2 to 2hr each of RGB, and 1 1/2 to 3hr of Ha or OIII, if
required; twice this much of each for really high quality images) . Don't forget flats (to
remove dust doughnuts and vignetting), biases are needed to calibrate the flats and lights
in prep for flat calibration, darks are necessary if your camera is noisy (go with the Sony
sensor to avoid the need for darks). Don't forget the light pollution (band cut) filter if you
have local lighting (if you live umpteen miles out in the boonies, you may be able to go w/o
this filter, otherwise leave it in your imaging train).
Camera Example
The SXV-H9 (Sony ICX285AL, anti blooming monochrome CCD) has low readout noise (6-7e
rms), requiring only a few minutes of exposure, under my (mag20/sec^2, typical zenithal
limiting mag of mid 5's) sky, to swamp it. Having said that, I try to expose long enough to
bring the background up to at least 2000 ADU (compression begins around 45-50K ADU,
unbinned on the SXV), without saturating the subjects highlights areas, or blowing out too
many "bright" stars (check the image statistics, especially peak and average, around the
subject, and the background). If the highlights dictate, I'll back off on the 5 to 10 minute
exposure that I use on typical subjects (galaxies and nebulas). For example, high surface
brightness nebulas, like (most) planetaries, require no more than a minute or two to loose
highlight details. Nebulas like M42 require at least a few minutes to bring out the
surrounding nebulousity, but the trapezium region will blow out in 15 or 20 seconds.
Compromise, use 1 minute exposures, or plan on a multiple exposure combo using PS, take
a complete sequence of LRGB for a few minutes, and another for a 1/2 minute. When
properly combined (layered), this will result in a multi-stop increase in dynamic range over
a single sequence. Narrow band filters like Ha will require longer exposures to bring up
the background, I generally expose for 80-100% longer than the clear filter exposure, but
keeping it less than 12 minutes to avoid the need for darks (the Sony chips used in the SXV
don't really benefit from darks until the exposure exceeds 12 or 15 minutes, run a 3x3
median filter on all subs (to remove hot pixels) and use SigClip or SD mask combine in
MaximDL. Take a test shot through the desired filter, use your image processing tools to
determine the background and highlight levels, increase, or decrease the exposure to get
the background up without blowing out the highlights.
Selecting the Camera/Scope Image Scale and FOV
I use the ED80/ f7.5 with the SXV-H9 (FOV 51' X 38') on large subjects, frequently with a focal
reducer (X0.7, more recently a TV X0.8) for an even larger field. This scope (and the ED100/
f9.0) offers good performance at a low price (you sometimes need a blue cut like that
offered by the Astronomik CLS filter, beware of the anti-fringe, they really cut the blue). The
ED100 (FOV 34' X 26') was occasionally used, but has since been sold, and SPX250/ f4.8 newt
(FOV 26' X 19') is most frequently used for galaxies, clusters, and nebulas with the SXV
camera. The Mewlon 250 (FOV 10' X 8') and SXV camera is used for "small" galaxies,
clusters, and all planetary nebulas. The Tak reducer/flattener provides very small spot
sizes from boresight to a 20mm radius, plenty for an APS-C sensor.
I attempt to select the scope that best contains the subject, leaving a some space around
the subject, you want to draw the viewer's eyes into the frame, not have them wandering
off to the sides, or out of the frame. Sometimes, on large subjects, this isn't possible, very
small subjects may have too much space around them. Frame the mass of the subject near
one of the 1/3 intersection points, or between them, sometimes this doesn't work well, but
generally does. Most of us are imaging for our own edification, not for scientific purposes.
If you are willing to free yourself by acknowledging that no qualified scientist would have
the least professional interest in your photos, then you can work toward your best
rendition of a "pretty picture". If the pretty picture is your goal, pay heed to (many) of the
guidelines that a century of photography has found to work.
Consider the image scale. My SXV-H9 has 6.5um pixels, which is typical (most cameras use
pixels that are 5 to 9um). This camera on the 80/f7.5 w/ a X0.8 reducer yields a scale of 2 3/4"
/pix. The same camera on my 10"/f4.8 newt yields 1.1"/pix. My seeing, which is typical,
ranges from 1.5 to 2.5". There is little point in running an image scale less than 1/2 seeing
(the losses exceed the gains), and you loose resolution if you go with a scale that is much
coarser than your seeing. If you are interested in capturing significant detail in planetary
nebula or galaxies (both can have a lot of interesting detail), use a focal length that allows
a scale on the order of 1" or 1.25"/pix. If you are imaging large nebulas, a scale on the
order of 2" to 2.5"/pix is adequate (the image will also be less sensitive to guiding and
seeing at the coarser scale). If you want true wide angle shots, get a camera with a larger
FOV (more pixels), but keep the image scale were it should be, in the range of 1" to 2.5"
per pix (I know, some people are getting decent looking camera lens shots at 5" or 10"/pix,
easy guiding, but no small scale detail). If you are using a long focal length, you may need
to bin the camera (2X2 is most common for monochrome CCDs, you can't bin one shot Bayer
Matrixed RGBs and retain the color info).
Guiding
A few comments about guiders. The real test of a guided system is the appearance of your
stars. Tight (seeing limited) round stars tells you it works. Misshapen (Oval) stars, or big fat
stars tells you something isn't right. If the ovals are all pointing in the same direction, it's
probably your guiding. You need to fix that before you worry about taking pictures of
anything. If the ovals seem to point away from some region in the photo, it's probably your
optics, there's too much curvature. You need to investigate "field flatteners". If you see
the stars becoming little comets towards the edge of the photo, take a look at "coma
correctors. Misshapen stars represent lost resolution. If you're giving up resolution, due to
guiding, or optics, fix it before you spend time worrying about photos.
The best way to avoid differential shifts and vibration, that can lead to misshapen stars, is
to use an OAG on the main OTA. The big problems with the OAG approach is finding
acceptable guide stars and framing the desired subject while keeping the guide star in the
guider FOV. You need to rotate the guider, around the LOS to find a star in the peripheral
field annulus, then independently rotate the imaging camera to reorient the subject,
sometimes translating both in RA/DEC to frame the subject while keeping the guide star in
the guider FOV. Sometimes there simply aren't any guide stars to be found, sometimes you
find one or two guide stars, but at the expense of moving the subject around to an
undesirable location in the imaging frame. Keep in mind that some OTAs (like Dall Kirkham
cassegrains) show significant curvature near the field edge. This results in guide stars
taking on noncircular shapes that change with variations in seeing (potential centroiding
issues).
I use a hard mounted, piggyback refractor on the ED80, TMB92, and 10" newt, it's SO much
more convenient than the OAG. Not better, just more convenient. If you decide to go with
the piggyback approach, remember the terms "differential motion" and "differential
flexure". You will probably see a slight movement in your stars, maybe a pixel in RA, maybe
declination, or both between sub-exposures. If you do, start tightening all the piggyback
scope mounting screws/bolts. If you use hard mounted tube rings, you will have less
problem than with the Losmandy rings, but you will give up the freedom of independently
pointing your guide scope. If you do choose the Losmandy approach, tighten those six tube
screws as tight as you think they should be against your guide scope tube, then give each
another (equal) partial turn. You'll see less image creep (some of that may also be due to
physical changes in the primary mirror if you use a reflector as your imaging tube. Be sure
your camera and filter attachments are also very well torqued down, tight. Many
inexpensive focusers are prone to deformation under gravity loads, keep the focuser in
mind when trying to track down an apparent differential motion problem. Long focal length
imaging benefits from use of the OAG, therefore, I use the Lumicon OAG with a coma
corrector (to get the guide star shape back toward something that begins to resemble a
circle) on the Mewlon, it's the only way to guide at 3000mm.
As for guiding, I guess that the ideally balanced, zero backlash mount is really the way to
go, but that isn't the CGE (the paramount ME may approach it). I find RA balance to be
relatively tolerant, there is a wide range of (mis)balance over which it operates properly, as
long as the load is well below the manufacturers suggested load limit (like 60% of said
limit). I put the RA weights so that it is slightly heavy on the east side (comes to balance,
with the scope at the desired DEC, at an elevation just below were I will begin imaging on
the east side of the meridian), the motors will be required to pull the load over the entire
range of RA that I'll be working. This requires resetting the weights when you flip to the
west. The DEC requires backend loading (what I consider to be a significant amount), and
the use of a little antibacklash compensation (Alt positive and negative typically set
between 5 - 10 on a scale of 0-100, typically not equal, using either the hand controller or
NexRemote on the laptop). I turn on the PEC before making the guider cal, and leaving it
running until I am ready to flip, or shut down for the night. PECtool (allowing multiple worm
cycle averaging) significantly reduces the PE, and large "noise" associated with the worm
imperfections (there is still residual "noise"). In order to gain maximum benefit, it is a good
idea to train it with 10 or more worm cycles. Once properly trained, it allows for more
aggressive guiding, with less error. My PE decreased from 18 arcsec p-p (uncorrected) to 7
arcsec p-p using PEC. The residual noise (sometimes referred to as asynchronous noise)
requires that the guider updates be made about every 1 to 1.2 second, 0.8 seconds is even
better, if the guide star s/n is high enough, and seeing is good. The MaximDL guider cal
routine is used to get the guide cam angle (I physically rotate the cam to result in
something between +/-5 degrees to keep the guiding DEC axis were it should be), I vary the
cal rates, and aggressiveness settings, to result in better guiding. I typically see about 1/3
arcsec rms on each axis, if seeing is OK. I've read claims of 1/4 arcsec on similar mounts.
I recall the saying..."In God we trust, all others bring data".
The CG5 mount requires a greater amount of (east side) RA and (backend) DEC loading. I
also use unidirectional DEC guiding on the CG5 in some cases. Guider updates are
required at intervals of 0.7 to 0.8 seconds, no PEC is offered for the CG5, so there are
significant short term variations from the desired RA tracking rate. The friction, backlash,
and stiction are appreciable on inexpensive mounts like the CG5, but they still work OK for
FL up to about 500 or 600mm under light loading (as with a couple relatively short FL, light
weight refractors) as long as the wind is "calm" (build a few 6'X8' wind screens out of 2"
PVC pipe and vinyl tarps, you may be amazed at how much better your mount guides when
there is no wind).
I can only say that the MI-250 is worth every penny. Without PEC, I typically see 4.5 to 5
arcsec of PE, and it's relatively smooth (much less "jumpy" than with the CGE). In other
words, the PEC isn't really necessary for guiding with this mount. This is typical of mid and
larger sized mounts, it's what you pay for. While PEC isn't abolutely necessary, I want to get
the most out of the mount, so I do, however, use PEC. I use PemPro to arrive at the proper
PEC "training" curve. With the proper set-up data, I see 2 arcsec peak-peak PE, and have
had slightly less (you'll have to make several PemPro runs before you see exactly what is
needed). The reduced PE allows me to run at least a few minutes w/o guiding. This can
make a difference if I just can't find a decent guide star. It also allows for longer guide
exposures, I typically run 2-4 secs, good when there aren't any bright guide stars (and it
helps average out the effects of "seeing").
Remember, misshapen stars point to reduced resolution, fix the problem before you worry
about photos.
Two stages encompass the image creation, photon collection, and post collection
processing. There are a few key items that result in good photon collection, select the
appropriate FOV (the right scope/lens FL for your camera), frame your subject as per the
photographic rule of thirds (there are exceptions), expose long enough to bring out the
background interests and swamp read noise (but not so long as to raise the sky
background too high or compress/saturate the scene highlights), select a sufficiently bright
(high s/n) guide star for reasonable guiding, properly set your mount guiding parameters,
and collect enough frames through each filter to yield a smooth field and allow stretching.
As for photographic rules, some apply, some don't. Remember that you are taking a picture
of some, specific subject, not just a chunk of the sky. Declutter...bring your subject to the
image foreground, push the clutter into the background. But everything is at the same
distance photographically, you say. Highlight your primary subject, de-emphasize the
unnecessary clutter, you know what your subject is, and what is cluttering up the photo.
Collect lots of photons through each of the filters that you are using (lum, chrome,
narrowband; 3 to 4 hrs of lum, 1 1/2 to 2hr each of RGB, and 1 1/2 to 3hr of Ha or OIII, if
required; twice this much of each for really high quality images) . Don't forget flats (to
remove dust doughnuts and vignetting), biases are needed to calibrate the flats and lights
in prep for flat calibration, darks are necessary if your camera is noisy (go with the Sony
sensor to avoid the need for darks). Don't forget the light pollution (band cut) filter if you
have local lighting (if you live umpteen miles out in the boonies, you may be able to go w/o
this filter, otherwise leave it in your imaging train).
Camera Example
The SXV-H9 (Sony ICX285AL, anti blooming monochrome CCD) has low readout noise (6-7e
rms), requiring only a few minutes of exposure, under my (mag20/sec^2, typical zenithal
limiting mag of mid 5's) sky, to swamp it. Having said that, I try to expose long enough to
bring the background up to at least 2000 ADU (compression begins around 45-50K ADU,
unbinned on the SXV), without saturating the subjects highlights areas, or blowing out too
many "bright" stars (check the image statistics, especially peak and average, around the
subject, and the background). If the highlights dictate, I'll back off on the 5 to 10 minute
exposure that I use on typical subjects (galaxies and nebulas). For example, high surface
brightness nebulas, like (most) planetaries, require no more than a minute or two to loose
highlight details. Nebulas like M42 require at least a few minutes to bring out the
surrounding nebulousity, but the trapezium region will blow out in 15 or 20 seconds.
Compromise, use 1 minute exposures, or plan on a multiple exposure combo using PS, take
a complete sequence of LRGB for a few minutes, and another for a 1/2 minute. When
properly combined (layered), this will result in a multi-stop increase in dynamic range over
a single sequence. Narrow band filters like Ha will require longer exposures to bring up
the background, I generally expose for 80-100% longer than the clear filter exposure, but
keeping it less than 12 minutes to avoid the need for darks (the Sony chips used in the SXV
don't really benefit from darks until the exposure exceeds 12 or 15 minutes, run a 3x3
median filter on all subs (to remove hot pixels) and use SigClip or SD mask combine in
MaximDL. Take a test shot through the desired filter, use your image processing tools to
determine the background and highlight levels, increase, or decrease the exposure to get
the background up without blowing out the highlights.
Selecting the Camera/Scope Image Scale and FOV
I use the ED80/ f7.5 with the SXV-H9 (FOV 51' X 38') on large subjects, frequently with a focal
reducer (X0.7, more recently a TV X0.8) for an even larger field. This scope (and the ED100/
f9.0) offers good performance at a low price (you sometimes need a blue cut like that
offered by the Astronomik CLS filter, beware of the anti-fringe, they really cut the blue). The
ED100 (FOV 34' X 26') was occasionally used, but has since been sold, and SPX250/ f4.8 newt
(FOV 26' X 19') is most frequently used for galaxies, clusters, and nebulas with the SXV
camera. The Mewlon 250 (FOV 10' X 8') and SXV camera is used for "small" galaxies,
clusters, and all planetary nebulas. The Tak reducer/flattener provides very small spot
sizes from boresight to a 20mm radius, plenty for an APS-C sensor.
I attempt to select the scope that best contains the subject, leaving a some space around
the subject, you want to draw the viewer's eyes into the frame, not have them wandering
off to the sides, or out of the frame. Sometimes, on large subjects, this isn't possible, very
small subjects may have too much space around them. Frame the mass of the subject near
one of the 1/3 intersection points, or between them, sometimes this doesn't work well, but
generally does. Most of us are imaging for our own edification, not for scientific purposes.
If you are willing to free yourself by acknowledging that no qualified scientist would have
the least professional interest in your photos, then you can work toward your best
rendition of a "pretty picture". If the pretty picture is your goal, pay heed to (many) of the
guidelines that a century of photography has found to work.
Consider the image scale. My SXV-H9 has 6.5um pixels, which is typical (most cameras use
pixels that are 5 to 9um). This camera on the 80/f7.5 w/ a X0.8 reducer yields a scale of 2 3/4"
/pix. The same camera on my 10"/f4.8 newt yields 1.1"/pix. My seeing, which is typical,
ranges from 1.5 to 2.5". There is little point in running an image scale less than 1/2 seeing
(the losses exceed the gains), and you loose resolution if you go with a scale that is much
coarser than your seeing. If you are interested in capturing significant detail in planetary
nebula or galaxies (both can have a lot of interesting detail), use a focal length that allows
a scale on the order of 1" or 1.25"/pix. If you are imaging large nebulas, a scale on the
order of 2" to 2.5"/pix is adequate (the image will also be less sensitive to guiding and
seeing at the coarser scale). If you want true wide angle shots, get a camera with a larger
FOV (more pixels), but keep the image scale were it should be, in the range of 1" to 2.5"
per pix (I know, some people are getting decent looking camera lens shots at 5" or 10"/pix,
easy guiding, but no small scale detail). If you are using a long focal length, you may need
to bin the camera (2X2 is most common for monochrome CCDs, you can't bin one shot Bayer
Matrixed RGBs and retain the color info).
Guiding
A few comments about guiders. The real test of a guided system is the appearance of your
stars. Tight (seeing limited) round stars tells you it works. Misshapen (Oval) stars, or big fat
stars tells you something isn't right. If the ovals are all pointing in the same direction, it's
probably your guiding. You need to fix that before you worry about taking pictures of
anything. If the ovals seem to point away from some region in the photo, it's probably your
optics, there's too much curvature. You need to investigate "field flatteners". If you see
the stars becoming little comets towards the edge of the photo, take a look at "coma
correctors. Misshapen stars represent lost resolution. If you're giving up resolution, due to
guiding, or optics, fix it before you spend time worrying about photos.
The best way to avoid differential shifts and vibration, that can lead to misshapen stars, is
to use an OAG on the main OTA. The big problems with the OAG approach is finding
acceptable guide stars and framing the desired subject while keeping the guide star in the
guider FOV. You need to rotate the guider, around the LOS to find a star in the peripheral
field annulus, then independently rotate the imaging camera to reorient the subject,
sometimes translating both in RA/DEC to frame the subject while keeping the guide star in
the guider FOV. Sometimes there simply aren't any guide stars to be found, sometimes you
find one or two guide stars, but at the expense of moving the subject around to an
undesirable location in the imaging frame. Keep in mind that some OTAs (like Dall Kirkham
cassegrains) show significant curvature near the field edge. This results in guide stars
taking on noncircular shapes that change with variations in seeing (potential centroiding
issues).
I use a hard mounted, piggyback refractor on the ED80, TMB92, and 10" newt, it's SO much
more convenient than the OAG. Not better, just more convenient. If you decide to go with
the piggyback approach, remember the terms "differential motion" and "differential
flexure". You will probably see a slight movement in your stars, maybe a pixel in RA, maybe
declination, or both between sub-exposures. If you do, start tightening all the piggyback
scope mounting screws/bolts. If you use hard mounted tube rings, you will have less
problem than with the Losmandy rings, but you will give up the freedom of independently
pointing your guide scope. If you do choose the Losmandy approach, tighten those six tube
screws as tight as you think they should be against your guide scope tube, then give each
another (equal) partial turn. You'll see less image creep (some of that may also be due to
physical changes in the primary mirror if you use a reflector as your imaging tube. Be sure
your camera and filter attachments are also very well torqued down, tight. Many
inexpensive focusers are prone to deformation under gravity loads, keep the focuser in
mind when trying to track down an apparent differential motion problem. Long focal length
imaging benefits from use of the OAG, therefore, I use the Lumicon OAG with a coma
corrector (to get the guide star shape back toward something that begins to resemble a
circle) on the Mewlon, it's the only way to guide at 3000mm.
As for guiding, I guess that the ideally balanced, zero backlash mount is really the way to
go, but that isn't the CGE (the paramount ME may approach it). I find RA balance to be
relatively tolerant, there is a wide range of (mis)balance over which it operates properly, as
long as the load is well below the manufacturers suggested load limit (like 60% of said
limit). I put the RA weights so that it is slightly heavy on the east side (comes to balance,
with the scope at the desired DEC, at an elevation just below were I will begin imaging on
the east side of the meridian), the motors will be required to pull the load over the entire
range of RA that I'll be working. This requires resetting the weights when you flip to the
west. The DEC requires backend loading (what I consider to be a significant amount), and
the use of a little antibacklash compensation (Alt positive and negative typically set
between 5 - 10 on a scale of 0-100, typically not equal, using either the hand controller or
NexRemote on the laptop). I turn on the PEC before making the guider cal, and leaving it
running until I am ready to flip, or shut down for the night. PECtool (allowing multiple worm
cycle averaging) significantly reduces the PE, and large "noise" associated with the worm
imperfections (there is still residual "noise"). In order to gain maximum benefit, it is a good
idea to train it with 10 or more worm cycles. Once properly trained, it allows for more
aggressive guiding, with less error. My PE decreased from 18 arcsec p-p (uncorrected) to 7
arcsec p-p using PEC. The residual noise (sometimes referred to as asynchronous noise)
requires that the guider updates be made about every 1 to 1.2 second, 0.8 seconds is even
better, if the guide star s/n is high enough, and seeing is good. The MaximDL guider cal
routine is used to get the guide cam angle (I physically rotate the cam to result in
something between +/-5 degrees to keep the guiding DEC axis were it should be), I vary the
cal rates, and aggressiveness settings, to result in better guiding. I typically see about 1/3
arcsec rms on each axis, if seeing is OK. I've read claims of 1/4 arcsec on similar mounts.
I recall the saying..."In God we trust, all others bring data".
The CG5 mount requires a greater amount of (east side) RA and (backend) DEC loading. I
also use unidirectional DEC guiding on the CG5 in some cases. Guider updates are
required at intervals of 0.7 to 0.8 seconds, no PEC is offered for the CG5, so there are
significant short term variations from the desired RA tracking rate. The friction, backlash,
and stiction are appreciable on inexpensive mounts like the CG5, but they still work OK for
FL up to about 500 or 600mm under light loading (as with a couple relatively short FL, light
weight refractors) as long as the wind is "calm" (build a few 6'X8' wind screens out of 2"
PVC pipe and vinyl tarps, you may be amazed at how much better your mount guides when
there is no wind).
I can only say that the MI-250 is worth every penny. Without PEC, I typically see 4.5 to 5
arcsec of PE, and it's relatively smooth (much less "jumpy" than with the CGE). In other
words, the PEC isn't really necessary for guiding with this mount. This is typical of mid and
larger sized mounts, it's what you pay for. While PEC isn't abolutely necessary, I want to get
the most out of the mount, so I do, however, use PEC. I use PemPro to arrive at the proper
PEC "training" curve. With the proper set-up data, I see 2 arcsec peak-peak PE, and have
had slightly less (you'll have to make several PemPro runs before you see exactly what is
needed). The reduced PE allows me to run at least a few minutes w/o guiding. This can
make a difference if I just can't find a decent guide star. It also allows for longer guide
exposures, I typically run 2-4 secs, good when there aren't any bright guide stars (and it
helps average out the effects of "seeing").
Remember, misshapen stars point to reduced resolution, fix the problem before you worry
about photos.
