Capturing "Galaxyset" - Paul Schmit

Capturing "Galaxyset"

Hello everyone!  This is my first-ever how-to tutorial, where I plan to review the entire process I used to obtain one of my new favorite gallery images.  In this tutorial, I will describe how I planned, executed, and post-processed Galaxyset, which depicts our nearest galactic neighbor, Andromeda, and her sister satellite galaxies setting over the picturesque Engineer Mountain in southern Colorado.  Nearly every aspect of this image was premeditated, and many useful applications, websites, and associated thought processes lended themselves to eliminating the biggest uncertainties from the shoot.

Truth be told, in anticipation of the inevitable conclusion to the Milky Way core photography season in late October, I started writing down ideas for other astrolandscape compositions I'd like to attempt in the "off season."  Knowing that Andromeda was a sizable (nearly 6x larger than the moon in apparent angular size!), albeit dim, object in the night sky, it seemed like a perfect candidate to work into an astrolandscape at a modest telephoto focal length.  Being a resident of northern New Mexico, my initial instinct was to attempt to capture Andromeda setting over Mt. Wheeler, New Mexico's tallest peak.  How cool would that be???  However, lots of subsequent intermittent research and planning led to a complete overhaul of my original, naive plans, and I will discuss the biggest challenges associated with capturing a galaxyset in what follows.

Fig. 1: Andromeda's trajectory in the night sky

Right Place, Right Time

Capturing an image of this nature isn't just a lucky accident.  To be successful, we have to take several deliberate steps to ensure that we will be in exactly the right place, at precisely the right time, to get our shot.  To begin, for those running iOS on their phones/tablets, I highly recommend the app, Sky Safari Pro, to help plan your astrophotography exploits.  Figure 1 shows a sequence from Sky Safari depicting the position of Andromeda in the minutes just before and after astronomical twilight the morning I captured the image, just one day before the new moon, prime time for photographing deep-sky objects.

This data sets some very important constraints on a successful astrolandscape composition.  First, in the upper left corner of each frame, we see that the elevation of Andromeda prior to astronomical twilight is in the neighborhood of 12-18 degrees above the horizon.  So, to capture Andromeda before the sun washes it out, it won't actually be setting below the horizon; rather, I have to create my own galaxyset by playing with perspective.  Thus, my foreground element (say, a stunning mountain peak) needs to reach up toward a similar elevation angle to be positioned closely to Andromeda prior to morning twilight.  This requires identifying a candidate mountain, and then figuring out how far I need to be from it so its peak sits 15 degrees above the horizon.  To make things simple, here's a quick formula to get right to the answer: if a mountain peak sits at an elevation H (in feet) above its surroundings, we need to position ourselves a distance D (in feet) away from the peak in order for the peak to sit at the desired angle A (in degrees) above the horizon, where D=57 x H / A.  For example, Engineer Peak sits about 2700 feet above nearby highway 550, so I needed to position myself 57 x 2700 / 15 = 10,260 feet = 1.9 miles away from the peak.  (For the math-minded, the exact formula is D=H/tan(A x pi / 180), but the approximate formula will work fine when A<30 degrees.)

Another quick calculation here reveals that the angle of view at my preferred focal length range of 160-200mm maxes out at 8.6x12.8 degrees.  Since Sky Safari shows that Andromeda will be oriented vertically and spans about 3 degrees on its long axis, we see that Andromeda will occupy about ~3/8.6=0.35, or about a third of the whole frame, at this focal length!  Unfortunately, this also means that a landscape-oriented shot capturing some sky above Andromeda (to balance the composition) and below Andromeda (to capture enough foreground) will definitely not see the majority of the mountain, but really only the top ~4-5 degrees of it.  Thus, to plan my shot, it is also necessary to find a peak that is narrow/jagged enough that some interesting features are captured in this relatively narrow, telephoto field of view.  Incidentally, most of the tall peaks in New Mexico are also very broad, making them somewhat less compelling than the distinctively jagged peaks of southern Colorado for this particular shot.

Finally, our preliminary research sets one more EXTREMELY important constraint.  Just before morning twilight, Andromeda is set to be located at 310 degrees relative to due north (going clockwise around the compass), or in laymen's terms, to the northwest.  Thus, not only do we need to find a spot 1.9 miles from a 2700-foot-tall object (or 3/4 mile from a 1000-foot-tall object, etc.) that is sufficiently jagged/narrow to be visually interesting, but it also has to have road access to its southeast to afford a suitable view of the peak juxtaposed with our intergalactic neighbor.  Another long story short: after ruling out pretty much every peak in northern New Mexico, plus ruling out light polluted areas near Durango (here's a great tool to get a handle on light pollution!), I happened upon a perfect opportunity with Engineer Peak, about 15 minutes north of the Purgatory ski resort.

Fig. 2: The amazingness that is TPE

Fig. 3: TPE 3D scene rendering vs. actual shot

Bringing It All Together

My hunt to find an appropriate foreground started with hours of intermittent browsing on Google Earth, looking at tall peaks and the road networks surrounding them.  But, the tool that turned an idea into a concrete plan was The Photographer's Ephemeris app suite.  Some other folks have found tremendous success with PhotoPills for similar applications, but I started with TPE, have racked up a number of successful shoots, and have mostly enjoyed the user experience.  (I plan to try PhotoPills soon for some upcoming astrolandscapes going into the new year to compare to the TPE user interface and functionality.)

The core component of my workflow begins with identifying a candidate peak, and placing a "secondary" (gray) pin at the location of the peak, signifying the prospective subject of my composition.  Then, with my secondary pin placed, I began moving the "primary" (red) pin around to various spots southeast of the peak to see if a suitable vantage point exists to view Andromeda setting.  Lo and behold, when it comes to Engineer Mountain, a point right along the very aptly placed Highway 550, depicted by the first screenshot in Fig. 2, fits the bill: it sits about 1.9 miles from the peak (satisfying the ~15-degree angle-of-elevation constraint on our desired composition, verified in the "A: +15.7 degrees" metric displayed in the lower half of the image), and it shows a bearing of 305 degrees to the gray secondary pin, within 5 degrees of the compass location of Andromeda just before morning twilight.  Additionally, the solid+dashed white line portrayed in the bottom half of the TPE screenshot shows the elevation change of the Earth's surface between the camera's vantage point (corresponding to the red primary pin) and the camera subject (corresponding to the gray secondary pin), with solid white lines indicating portions of the terrain that can be seen visually from the red pin location, and dashed parts likely being obscured from view by other parts of the local geography.  From the solid lines in the figure, we see that the top several degrees of the peak should be visible, as well as a substantial fraction of the lower-elevation ridgeline nearer the camera location.

Even better, referring to the second image in Fig. 2, a Google Maps screenshot showing two of my top potential vantage points, one of my top spots has a conveniently placed pullout along the otherwise tree-lined highway.  In fact, the existence of this pullout is the primary reason I risked the 9-hour round trip and overnight stay in Purgatory for the sole purpose of capturing this shot.  Without it, I had little confidence I could secure a viable shooting location along the highway with a suitable view of the mountain as well as Polaris (for polar alignment), and no obvious side roads seemed to be available, either.  Interestingly, despite these two marked spots being only a few hundred feet apart, going with the northern location (which is what I wound up having to do due to pervasive tree cover) forces me to a focal length of 160mm to capture the whole scene, while the southern spot captures the same scene at 200mm, further increasing the relative size of Andromeda.  So, this shot truly requires positioning accuracy to within a couple hundred feet to deliver the desired result!

Note: to illustrate the challenge in identifying this photo op, I've included in the bottom half of Fig. 2 a couple other examples of my TPE analysis for this shot for other prominent peaks: Wheeler Peak in NM, and Blanca Peak in CO.  In these cases, both peaks have complementary vantage points to capture galaxyset, each satisfying the elevation-angle and bearing requirements, but neither of these vantage points is easily accessible by road or trail.  Maybe next time...

Finally, to fine tune and finalize my location planning, I move over to the TPE 3D app with some candidate locations saved from my scouting in TPE.  Figure 3 depicts an incredibly useful screenshot from TPE 3D, showing my prospective view of the peak of Engineer Mountain on the morning of 11/6, just prior to morning twilight.  The app can faithfully represent your approximate field of view of 3D topographical features against the night sky at a variety of focal lengths.  Fig. 3 indicates that at ~160mm focal length, I should be able to visualize the entire peak of the mountain just as Andromeda glides into place, nestled between the northern slopes of the peak and a slanted ridge closer to the vantage point.  For Fig. 3, I've generated a slideshow juxtaposing this synthetic view against the final image to show that, while not perfect, this 3D representation was good enough to get me the shot I had planned for!  (I've also outlined the three guide stars I typically use to locate Andromeda in the night sky, as I can just barely see them in live view.)

Well, that's it!  We've planned our shot!  There's a phenomenally scenic peak, Engineer Mountain, with perfect road access at just the right distance and bearing to give us a compelling view of galaxyset before twilight.  All that's left to do now is head out to the location, capture the shots, and turn them into a piece of art!

Fig. 4: The view from our location

15 Degrees up and 15 Degrees Out

There!  Do you see it???  In Fig. 4, a sequence I took with my second camera while capturing Galaxyset with my primary camera setup (pictured on the right side of the images), we clearly see the faint, diffuse glow of Andromeda descending along the northern slope of Engineer Mountain, just as we predicted from all that preparation!  I captured this sequence with my Nikon D500 + Tamron 15-30mm f/2.8ultra-wide zoom (my main Milky Way astrolandscape lens) after my main photographic rig was set up and clicking away.  For what it's worth, my intention was to try to capture a buddy shot with my dad, who braved the marathon journey with me and suffered through 2.5 hours in the middle of the night in 15-degree weather so I could capture this shot.  In the end, I wound up with a pretty cool poor-man's time-lapse showing some of the critical moments of the shoot.

It's worth mentioning that I kept the entire week of 11/5 flexible as far as my work schedule was concerned, since any of those days would have been doable for capturing this image, with the new moon occurring on 11/7 and dark morning skies all week.  In the final days leading up to prime time, I followed the mid-range numerical weather forecasting models closely (primarily, the NAM and NAM NEST), hunting for days with the lowest cloud percentage forecasts over southwestern Colorado, along with the lowest humidity and winds at all elevations, particularly the lower half of the atmosphere.  My fear was not just typical cloudy skies (fog, low clouds, dense cirrus, etc.), but even localized, orographically induced lenticular clouds in otherwise clear skies caused by marginally saturated air flowing quickly over the mountain tops.  Although those would look spectacular in other contexts, the visible real estate north of the mountain through which I planned to view Andromeda was tight, and I needed every bit of clear sky I could get.  For some very helpful tools to peruse and analyze many common production weather forecasting models, check out this website.

My main photographic rig included a Nikon D810 full-frame camera body, a Nikkor 70-200mm f/2.8E telephoto lens, and an Orion Astroview German equatorial mount, which is a single-axis motor-driven mount intended mostly for use during casual observation sessions with telescopes, including the Orion 1300mm prime-focal-length f/13 Mak-Cas optical tube that I used to capture my ISS solar transit shots.  However, at modest telephoto focal lengths (approaching at least 200mm, which is the largest magnification I've tried), this mount has given me good tracking and sharp images for exposures approaching at least 1 minute, with good polar alignment.  I used a Vello wireless intervalometer to trigger my long exposures hands free (much cheaper than the brand-name equivalents and mostly straightforward/dependable). Also, to attempt to avoid potential issues with batteries getting cold and draining/malfunctioning, I used gaffer tape to tape some 10-hour hand warmers to the parts of my camera body and wireless receiver adjacent to their respective battery compartments.  Also, to avoid potential problems with fog/dew, I taped a few more hand warmers to the lens hood for my Nikkor lens, helping raise the temperature near the objective lens and keep it warmer than its surroundings, greatly assisting with fog/dew mitigation.

After we arrived at the target location, I first identified a spot offering me a glimpse of my desired composition between the tall conifer trees lining the highway.  Next, I set up my EQ mount and spent a good 30-40 minutes dialing in my polar alignment. I've confused myself before looking through the guide scope on the mount, which is surprisingly high magnification, not knowing if I was looking at Polaris or a star nearby.  To solve this problem, I brought my telescope with me and used the attached finder scope (lower magnification than the alignment guide scope), set my declination angle to 90 degrees (so that the finder scope and alignment guide scope are looking at the same spot in the sky), and used the finder scope to roughly center my crosshairs on Polaris.  Afterward, I pivoted to the guide scope to fine tune my alignment.  For this part of the process, I've found the iOS app PS Align Pro to be very useful.  It is simple to use and contains a surprisingly comprehensive database of polar alignment scope reticle designs, including the design used on the Astroview guide scope.

After achieving reasonable polar alignment, I attached my DSLR + telephoto lens combo to my EQ mount via this dovetail adapter.  Now, here's the cool part.  I spent so much time planning this composition, knowing exactly where to be in order to view Andromeda setting to the right of Engineer Mountain, that I saved myself critical minutes in having to location the precise location of the faint, diffuse galactic blob in the night sky.  Once my camera was mounted, with my mount properly polar aligned, I loosened the locks on my right ascension and declination axes and rotated my camera freely to set up the foreground composition I had originally envisioned, with the mountain peak on the left side of the frame and some open space to the right where Andromeda eventually would be found as it set.  (Note: I actually wanted to shift my composition slightly to the left to include more of the mountain, but it turns out a tree I hadn't previously noticed was in my line of sight, and by that time I had already chewed through more than 50% of my setup time polar aligning in this location.  With no time left to reposition and realign, I was forced to settle with a slightly different foreground.  For what it's worth, that darn tree is sitting just outside the lower left corner of the visible image sequence depicted below in Fig. 5, and you can see in Fig. 4 how narrow the slots between the tall trees were offering marginal visibility of the event.) With my foreground composition locked on prior to the arrival of Andromeda, I engaged the lock on the declination axis, and finding Andromeda was simply a matter of stepping backwards along my right ascension axis and taking some test shots until it appeared in my screen.  See, I knew eventually the sky would rotate the galaxy into the right side of my frame when viewing the scene from that exact position, which meant with a properly aligned EQ mount, I could rotate along my RA axis and eventually find the galaxy positioned precisely in the same part of the frame!  Between that fact and the use of a triangle of helpful guide stars located near the Andromeda constellation (mu-And, nu-And, 32-And) that demarcate the location of the galaxy, finding Andromeda in my camera was a breeze!

All that was left at this point was to capture the data necessary to compile into Galaxyset.  The plan was as follows: in the ~15-20 minutes before galaxyset, as Andromeda was descending toward the horizon and settling into its cozy corner between the mountain and a nearby ridgeline, I planned to capture a series of 40-second exposures with the tracking motor on my mount engaged, providing me sharp data of the sky and correspondingly blurry foreground elements as the landscape  intruded into the frame.  Then, at the moment the galaxy had descended to the point relative to the landscape that I desired for my final image--essentially, when the foreground returned to the state I had composed initially, sans Andromeda--I would switch off the tracking mount and immediately capture another several static images of the foreground, where the stars would now become notably streaked.  For the two image sequences (tracked and untracked) feeding into the final, processed image, I didn't touch the camera or lens, and all exposure settings, focal length, and focal point (infinity) were preserved, and all data was captured in a roughly 15-minute period.  The only difference between the sequences was tracked vs. untracked.  To illustrate exactly what the camera saw as it was capturing the tracked and untracked image sequences, Figure 5 is an animation of every single raw frame capture used to create the final Galaxyset image, showing the sharp, nearly static sky juxtaposed with a blurry, advancing foreground (indicative of the tracker being on), followed by stationary images of the foreground with blurry/streaked stars + Andromeda (indicative of the tracker being off).  Note, I've added +2 EV to these images to make for easier viewing, but otherwise this is raw, unedited data.  Finally, with my sky and foreground data in hand, my mission was complete, and I packed up and loaded my gear back into my Subaru with numb fingers and toes and set back to the hotel for some much-needed shuteye prior to checkout and the 4+ hour drive back to Albuquerque.

QUICK EXIF NOTE: All exposures were 40s, ISO 800, f/3.5, 160mm.

Fig. 5: Raw, SOOC images (+2 EV) used to create Galaxyset

Processing Galaxyset

I will finalize this tutorial with a brief description of the post-processing techniques I used to convert the raw image data into the final "Galaxyset" image shown at the top of the page.  The first thing I did was convert all the raw files of my tracked sky shots to 16-bit TIFF files in Adobe Lightroom (LR).  Then, I used the Deep Sky Stacker (DSS) program to align the 18 sequential tracked images I took and output the aligned files as another set of TIFFs.  The remainder of my editing was done exclusively in Photoshop (PS).

Figure 6 shows a sequence of screenshots taken during several stages of the PS editing process.  The first frame shows the result of performing a median stack of the 16 best frames chosen by DSS after importing each 16-bit TIFF file as a layer and converting the 16 layers to a Smart Object.  (Note that without the +2 EV shift applied to the raw images in Fig. 5, this version looks pretty dark. But don't worry, the data is there!). In the next frame of the editing sequence, I've additionally loaded in my 6 static frames of the foreground, converted them to their own Smart Object, executed a median stack on the foreground frames, and applied a layer mask to blend the sky and foreground frames together.  For this step, I found the "Select-->Color Range..." selection tool quite useful for a first stab at selecting only foreground objects.  With a little cleanup using the brush tool in the "Select and Mask" environment, I had a pretty-near pixel-level mask revealing my foreground against the sky, leaving everything quite sharp and ready for the final editing steps.

Finally, the last three images in the editing sequence show the result of several applications of ACR filters to bring up the overall exposure level of the composite image, stretch the sky histogram to bring out faint details of Andromeda, and finally play around with the global hue and saturation levels.  I also used a star reduction technique similar to this to help draw the eye toward Andromeda more easily.  Interestingly, unlike previous Andromeda shots I've taken when it was nearly overhead, revealing quite a bit of subtle coloration in the galactic structure, these images came out biased much more toward the yellow side (despite identical white balance settings), likely due to atmospheric filtration of some wavelengths preferentially over others with Andromeda sitting this low in the sky, just like the coloration we might expect to see at sunset.  As a result, I spent a fair amount of effort trying to bring the white balance back to something neutral, and I almost made a decision to convert to monochrome to avoid the problem altogether.  However, the winning formula for me eventually turned out to be backing off the natural saturation of the image to partially wash out some of the rich foreground color and get to around "half monochrome," then adding a couple more color balance and hue/saturation adjustment layers to bias the desaturated image toward the blues, leading to a moody, surreal palette that captured the ambiance of the environment quite nicely.   The original image was captured in a 3:2 aspect ratio on my full-frame sensor, but I chose to crop Galaxyset to 16:9 for optimal viewing on digital devices.

Well, that's it folks, the making of Galaxyset from start to finish!  It was a pretty serious effort, and wasn't without its hiccups, but I'm quite pleased with the result and plan to improve on these methods more in the coming months.  Thanks for reading!

Fig. 6: Editing stages in Photoshop

Powered by SmugMug Log In