Tag Archives: night

The Elusive Milky Way – Capture an Image

Published: July 7, 2012
Last Updated: September 10, 2018

I assume you already read part one of this article which describes a bit about what the Milky Way is and what times and seasons are best for photographing the cloud-like expanse of innumerable stars.  In this installment we describe the equipment and settings you will need.

Just Ahead: A Universe of Possibilities

f/2.8, ISO 3200, 30 seconds, 16mm, post processed and combined with shots of the bridge that were lit with a spotlight.

Standard Capture

To get a passable or better image of the rather dim Milky Way you need:

  • A high performing low light camera (more on that in a moment)
  • A large aperture (f/2.8)
  • A wide angle lens. Ultra wide even.
  • A cool/cold night
  • As little city glow and moonlight* as possible – see below for an image taken in twilight
  • A solid tripod
  • Patience
  • To know where and when to look!

To get a recognizable Milky Way in a single frame, you’ll want to use somewhere between 2000 and 6400 ISO at f/2.8 or wider setting. That’s very high, and a wider aperture than many people have paid for.  You’ll also want to expose as long as you can before stars are streaking.  We recommend starting at 30 seconds, and reducing your exposure time if the streaking is objectionable. Below is an image taken when the rising moon was beginning to wash out the sky and this may be typical of attempting to capture the Milky Way in a less than ideally dark scenario. Just want a quick suggestion for settings:  Use these:

  • f/2.0; 24mm; ISO 6400; 15 seconds or
  • f/2.8; 24mm; ISO 3200; 25 seconds (or longer)
Group Hug

Moonlight and Twilight begin to overwhelm the Milky Way in Alabama Hills, California; 30 seconds, ISO 3200, f/2.8, 17mm

Some image degradation is to expected. For example vignetting and coma are both more obvious at lower f/stops. Coma is a comma or “bird-wing” like appearance of stars near the corners of the image.  Both coma and vignetting can be overcome by stopping down the shot – but resist the temptation because stopping down means losing some or perhaps all of the wispy milky goodness that you are trying to capture. Exposing longer will only help if you have some special apparatus (see Tracked Capture below). Are you wondering why exposing longer does not solve the problem? We have tackled the issue in two different styles: a cheerful allegorical example, and a recent math savvy explication.

What will an image look like captured with 3200 ISO? It may look like the image on the left below which is “straight out of the camera” – but perhaps not for you as this image was taken in a VERY dark sky area in Nevada.  On the right is the same Milky Way with some simple processing we will describe in the next installment.

SOOTC (and not SOOTC) [C_039467]

What is a “High Performing” Camera?

I qualified my statement earlier by indicating a high performing camera is needed for a standard capture like those I’ve shown above.  Since it would be impossible to keep an up-to-date list of the current high performing cameras, let me instead point out a few characteristics common to all high performers:

  1. Recent generation (2 or 3 years since introduction) is preferable because technology has steadily improved.
  2. Large pixels (to collect more light).  A common measure of the pixel size is in microns. Generally this puts full frame cameras ahead of cropped cameras.
  3. High “ISO at Unity Gain” – this is a measurement of the efficiency of the sensor. There are two good sources for this information: the DxO Sensor Scores and ClarkVision’s (older) tables.
Don’t be fooled by the highest ISO setting advertised. That number is completely meaningless.
As of August 13, 2018, the highest performers are listed by manufacturer and in order of performance. E.g. the Nikon D3s is better than the D800 – though the difference is small. Indeed, the D800 excels in some categories over the D3s. Cropped cameras are shown in italics – note that there fewer of them and none of the crop cameras exceed their full frame siblings. The first paragraph are the TOP performers. The next bracket list other cameras that “meet” our judgement of “good enough to photograph the Milky Way – with an appropriate lens. Note that the Cybershot DSC-RX1R ranks right after the Canon 1DX II – that’s quite a surprise –  it does have a fixed focal length of 35 mm, however.

TOP PERFORMERS

Pentax: 645Z
Hasselblad: X1D-50c
Sony: A7 III, A7S, A7R III, A9, A7R II, (Cybershot DSC-RX1R II – 35mm f/2.0 lens, A7S II)
Nikon: Df, D3s
Canon: 1Dx II

DECENT PERFORMERS

Nikon: D4s, D600, D800E, D4, D750, D610, D800, D810, D850, D5, D700, D3, D3X, D3300, D5200, D7100, D5100, D7000, CoolPix A, D3200
Canon: 1DX II, 5D IV, 6D II, 1Dx, 6D, 5D Mark III, 5D II, 1DS III, 1DS II, 5D, 1D III, 1D VI, 1D III, 1 D II
Sony: A7R, DSC-RX1R, RX1, A7, Alpha 99, Alpha 900, Alpha 850, A6000, Alpha 580, NEX-F3, NEX-C3, NEX-5N, NEX-3N, NEX-6, NEX-7
Leica: M Typ 240, X Vario
Phase One: P40 Plus, P65 Plus
Pentax: K-1, 645D, K-5 II, K-5 IIS, K5, K-50, K-01, K-30
FujiFilm: FinePix X100

Not in contention: any cameras by: Casio, Konica Minolta, Mamiya, Nokia, Olympus, Panasonic, Ricoh, or Sigma.

The list above shows all cameras having a DxO Sports (low light) score of 1000 or higher.

Cameras like the Nikon D90, Canon 1D II N, Phase One IQ 180, Canon 1Ds, Nikon D3100 and Leica M9 fall just below this threshold and may also be suitable.  The first eight Nikon models outperform the Canon 1Dx, and after the 1DX is the Sony A7R. The Fujifilm just barely cracks the list in 43rd and last place.

If you want the camera to cost less than $2,000 USD your current top choices are: Sony A7 III, Pentax K1, Nikon D610, Canon 6D II (or 6D).  If we were to make a recommendation, we’d recommend any of the full frame choices over the smaller sensor cameras.  Note that prices vary dramatically, and you may find used higher performing cameras for less than $2000. Beware of all Sony models, however, as they have had a long standing problem with “Star Eater” noise reduction problems. As of August 13, 2018, it’s not clear if they’ve actually fixed this problem on all of their models.

Stacked Capture

A “stacked” capture is what you may need to resort to if your camera performance is not so spiffy.  The approach applies astrophotography techniques to create a lower-noise version of an image.  The technique requires MANY shots of the same view. However using this approach you will want to avoid having anything but sky in your photo. Terrestrial elements will make stacking the image tricky.

Urban Milky Way [C_036919-23PSavg]The image at the left is a stacked capture to illustrate the point, however it was done with a high performing camera and only 5 images.  A lower performing camera will require as many as 20 or so captures to combat the noise. The method is described in my a “Astrophotography 101” Webinar and details are walked through in Astrophotography 301.  On the other hand, this image was captured in a location where the Milky Way was quite faint – alongside 7 million people in the San Francisco Bay Area so there is hope even where the Milky Way can only faintly be seen.

Details about the stacking method appeared in an earlier column as well as in an an earlier webinar.

Tracked Capture

The last way to get a great shot of the Milky Way is to track the sky with an apparatus called an Equatorial Mount.  By tracking the sky at the rate of the earth’s rotation you can lengthen a 20 second capture to perhaps a 60 second one. You can also use several such captures to create a stunning “Stacked Capture”. Again, however, shots which include the land are a bit harder to pull off unless you resort to layering. What do you need to do a tracked capture? We cover that in detail in the Astrophotography 101 Webinar, but in short, you’ll want an Equatorial Mount of some sort – not an Altitude-Azimuth (aka Alt-Az) mount! A device that looks intriguing and not terribly expensive is the Polarie.

Once you get that image (or those images), you will no doubt want to tease the most pleasing photo you can out of your data. That is a topic we’ll cover in the next installment: Processing your Milky Way images.

Skies Ablaze – Capturing Fireworks

Pod People [C_033010]

Photo 1: Abstract Sausalito fireworks (aka Pod People)

Several times a year Americans (and others around the world) find an excuse to gratuitously burn large amounts of black powder in spectacular displays that serve no real purpose except to create awe and wonder. For that reason alone, I love fireworks.

Fireworks are an interesting subject for Night Photography and require a little patience and experimentation to do well. And I have found some new approaches to photographing fireworks that make them even more interesting – as shown in Photo 1, above. Yes… that is a firework burst!  For information on how I achieved that effect, see below.

Challenges with Fireworks

Fireworks come with their own set of issues. From the ephemeral nature of the light to the selection of a site to photograph them from.  I generally favor more distant locations where I am able to switch between telephoto and wide angle lenses to alter the composition of the shot.  At greater distances the flashes of the fireworks are also a little less challenging photographically.

Brief Intense Light

Firework bursts are brief lasting seconds or less, but surprisingly they are very bright. Because they are so brief catching them at just the right moment may at first seem daunting. As with all things night photography the camera is quite inadequate at metering or adjusting for fireworks in a way that will capture the drama and grace. So the first tip:

Tip 1: Stay in MANUAL mode and do not let the camera try to adjust anything.

I generally prefer settings that are approximately like this:  200 ISO, f/9, 10 seconds.  But may change those dramatically based on the outcome of a shot or two, and how much background I’m trying to grab.  For example, I will often choose shorter exposures at higher ISOs, and sometimes I use BULB mode.  More on that in a moment.

Location, Location, Location

As in real estate, location is everything. Fireworks, in my opinion look better against a city skyline, reflected in water, or next to the moon or a snowy mountain peak. But creating an exposure to include those background elements is tricky, sometimes very tricky.

Nearly every event in which fireworks are employed is massively crowded (think 4th of July) so finding a clear view and a safe hassle free space to set up a tripod takes a little creativity and patience. Fortunately fireworks can be viewed from near or far – and far is often a bit better.

Photo 2: Fireworks and the San Francisco Skyline

Weather may also be problematic – as in the San Francisco area where fog in the summer evenings and mornings is normal. Four years of shooting produced only one year with moderate haze (as in Photo 2). Two years were complete shutouts.  Photo 2 could be improved with a brighter background – either by taking a longer exposure at the time or by combining with an exposure taken earlier or later in the evening.

The minions in the Night Photography group were clamoring to shoot fireworks, but thrice burned equals twice shy despite a fantastic location found years earlier.

Tip 2: A good location is always the better choice. Location, location, location!

Composition

Wide shot or tight shot?

Photo 3: Tight coverage of a burst

Photo 4: Wide angle with pond and reflection

Straight shot or creative shot:

Photo 5: Changing focus during exposure

Fortunately there is no rule that says you can not try a number of different approaches in the same event – provided the event lasts long enough.  However it is best to start with trying to get a pleasing result that is conventional in nature that is, more like Photo 4, above.

Timing

As in comedy, timing is important.  Start the exposure too soon and you’ll catch the bright burst of the firework which may very well overwhelm the camera and produce a “white blob”. There are three approaches to the timing problem, and sometimes I use all three in the same night:

  1. Set the camera up with an intervalometer/locked remote shutter and just let it run.
  2. Manually release the shutter using a fixed exposure length as soon as you see a burst.
  3. Manually control the shutter in bulb mode.

The main reason to use a fixed exposure length is if you intend to stack or animate your photos – keeping the background exposures the same produces a more pleasing, flicker free result. However if your only goal is to “get the boom”, option 3 is probably the best.  I recommend shooting after the initial “boom” because what makes the firework appealing is the light spreading and then falling over time.  How long to expose depends on a lot of things, of course, including how bright the background is and how many fireworks are blowing up at a time.  The Finale is usually a good time to take your eye off the camera and enjoy the show because finale’s usually end up a washed out mess.

Tip 3: Start the exposure immediately AFTER the boom.

Creative License

So how did I get the strange effects in Photo 1 and Photo 5?  I adjusted the focus while exposing.  I have also played with the zoom while exposing.  In Photo 1 the shot started focused and I defocussed it.  In shot 5 it was the other way around – it started out of focus and finished less out of focus.

Once you start playing with the focus, do not expect to have sharp images which is why I recommend you get the conventional exposure first before you get all creative.

 

Creating a Night Panorama

Asylum at the Sea

That’s it. You just found the perfect composition but there is a problem. The interesting bit does not fit with the other interesting bits.  So you sacrifice and change to a wider angle lens but that does not improve the shot. In fact, you can’t get an image wide enough or tall enough. Let’s change our frame of reference for a moment because not many people think about the ability of panoramas to connect the foreground to the heavens.

Although taking truly panoramic star circles is next to impossible within a reasonable budget, you can connect the earth with the sky with a little bit of planning and some tricks to aid in your alignment later in the process.

We are proposing a vertorama that is a “vertical panorama” to extend our photo to the ground and blend it into the star trail on top.  Our approach will be to shoot in landscape mode and tilt our camera up as we progress.   We can take a series of two or three shots but it will look odd if the star trails don’t extend into the edges of solid land (lower portions).  We also have to make sure that we don’t end the last frame too high as this leaves the star trail disconnected from the ground (lower) images.

Gear

The minimum amount of gear required is a camera a tripod and an interval timer.  However the whole process is going to get much easier if we have a method to align the “no parallax point” with the axis of rotation.  Huh? What did you just say…If you didn’t get that last sentence I suggest you read our primer articles on panoramas here and here or by following links out to other resources.

Planning a Shot

By this time I am assuming that you have read all of the prior articles on shot planning and alignment.  If not they can be found here, here, and here.  Now that we are up to date on planning and alignment we can get into the gory details. We are going to have to take some additional steps to help in the alignment in post processing.

I can get obsessive about the details but I want to do it right not get back to my computer to discover I spent 4 hours in the cold, dark night with stars and nothing interesting in the foreground.  I begin with an idea of how I want the image to look then I walk around with a compass and the local north declination adjustment to fine tune where the center of rotation will be in my final image. I set up my tripod where my chosen foreground will be in  alignment with Polaris (aka The North Star). Before I take the first photo I know where Polaris is going to appear in my frame. When I am really particular I will use a laser line to insure I have the object exactly where I want it in the photo.  I then level the tripod and assemble the pano head.  I use the pano head to determine the elevation of the north star this helps me know exactly how high the star with be in the sky.  This helps me to see where the star circle is going to be in the photo.  [Editor’s comment: If you know the angle of view of your lens you can determine the altitude of the north star by using your camera’s field of view as a measurement]. Knowing how high Polaris is in the sky will help me to determine the overlap I need to include all of the elements in the photo I want.  This all may seem like over-kill but to go home and thinking you have an image of a lifetime only to find you have a dud is not fun.

Shooting (and Bracketing)

I take a lot of photos with different settings until the blue period ends and the stars begin to appear. I make sure that I take a lot of bracketed shots that have 30% to 50%  of overlap. Having plenty for foreground images to choose from later is going to be very helpful when merging the final images.   Once I am satisfied I have enough images the next step is to lock the camera and tripod down tight to take the series of star trail photos.  If I didn’t get to my intended location in time and still want to get foreground shots it is still possible. Darkness merely means that my foreground exposures are going to have to be long, perhaps very long. Alternatively I can light the foreground with a flash light, strobe, or fire or I will let you borrow some light from my moon. 🙂

These are some of the images the top and bottom of the vertorama. I took right before the blue period ended. You can see the various bracketing and over lapping I did.

 

Post Process

I assume that you already know how to process your star shots into star trails. If you need a refresher check out Steven’s articles on star trail creation.  I like to do a few different versions of the with stars because I like to include different amounts of blue and so I have an easier time blending them in later in the stitching process.

Just some of the Star Circles I generated using different amounts of blue period photos.Combining a star circle with a foreground can be done with several different stitching engines, I prefer PTgui, however for other stitching engines the workflow is similar. Your images may be bright enough to do the whole process with Photoshop’s stitching or in Microsoft ICE engine but if you are blending in photos to get a darker sky then Photoshop might be a dud.  Also these stitching programs will not allow you to add control points to help align and warp the image to the background and let me be the first to  say the alignment and warping the foreground image is not fun or easy.

Import your Images

I like to import a lot of different bracketed images into the stitching engine just so I can have some variety to work with.  Also it will allow me the automatically find control points on overlapping images and then tell the software that the other images have zero pixel shift from that image in the bracket series.  This allows me to save time by putting control points on only the images that are dark or match the star circle image.   Thus I can align and blend the pano in one step then use Photoshop to blend in the stars.

Once imported click the align images,  did it work…if so hooray!! You now only have a small amount of clean up to do (skip down to the projection part of the article).  If not then you have to add some control points and align the images yourself.  Control points are areas on an image that match in overlapping images the software needs to know these areas to know how to warp adjacent images to stitch them together.  So lets look at how PTgui does this.  Below is a screen shot of PTgui’s control point placement feature.

Control points selected between the star circle image and the bright forground image. I pick areas in the image that have 1) sharp contrast, 2) Jagged edges and 3) don’t move.

All you have to do is open the adjacent images at the seam and then zoom in to find distinct points that match.  I like to use object with good contrast and unique shapes to help guide my cursor to a pixel accurate match.

In the image above you can see that the transition between the rock and the sky is the area of overlap.  I follow the rock edge because of the contrast between the sky and the rock but also it has bumps that are easily distinguishable between the top and the bottom photos. Once I have these set I do the same for the star trail photo see below.

Aligning the Star trail image with the photos from the blue period

I will align the star circle photo with the corresponding photo of my foreground just so I have less stitching errors and it is easier to align.  Again if the star circle is light enough then these images may automatically align.

Projection and orientation

Once all of the images have control points then I will go and see how the preview of the image looks in the panorama editor.

Looking at the stiched images in the panorama editor for the first time to see the errors in overlap and the corrections needed to correct the distortion.

You can see the top and bottom images of the vertorama are properly aligned at this point don’t worry about the blending we will handle that later in photoshop.  Two things are very off, causing distortion of the star circle 1) the projection and the vertical height of the image.  The projection is the way the images are projected on to the inside of a sphere during the alignment.  If you are interested in the types of projections  (and there are many in the image below) and how the distortion affects the overlap and warping of your image more info can be found here. I am not interested in the type or how it works in this case I am just worried about what looks right.  For most vertorama star circle images the “Rectilinear” projection often looks the best but is not always the case.

Correcting the projetion to make the photo appear flat and not as distorted at the edges.

We can quickly correct the oblong star circle by moving the top image toward the bottom. This will change the amount of space that is blank but once we crop the image those areas will not longer be visible anyway.

Changed to rectilinear and moved the whole image down. You can see I changed the vertical FOV with the slider on the right (right red arrow) I changed the projection at the top to rectilinear. and the line in the middle shows the movement of the image down.

 Outputting the files

Once you are satisfied with the preview the next step is to use the image optimizer to hone the control points to reduce the error.  It is so easy I am not even going to include a photo just click and tab and click optimize.  It will give you a rating “very good”, “good”, “not bad”, etc.  then suggest some corrections my usual experience is just accept them they more often help then harm.

The only step left is the output the files and then mask them in photoshop.  Go to the “Create Panorama” tab and preview the settings.  You can see the settings I use is to output the images as a “.psb” for use in photoshop but you can output what ever you would like.  The most important is to output into 16bit layers under the “LDR file format” or “HDR file format” settings. This is shown in the screen shot below.   Since the blending is going to  cancel out some of the stars the aligned images need to be output in seperate files so.  I will uncheck the image that represents the aligned star circle in the screen shot below this is “Image 6” the rest of the images are the aligned HDR brackets.  Then I will name the file something like “file….SC_BKGD” were file is the former file name and the “SC_BKGD” stands for “Star circle Background”.

Outputing the top and bottom layers

Once this is done stitching I output the star circle image alone.  So in this example I would un-check the boxes next to images 0-5 and only output image 6.   So once all of these images are outputted I select them and open them as layers in photo shop.  If your blending went well in PTgui then  great skip the next step and forward on to the next.  If the blending shows seams or other artifacts of stitching follow then next steps.

If the blending was bad then go back to PTgui mask out the areas the blending did not work so well if you have PTgui version 9, if not then output all of the layers and we will mask and clone in photoshop.  I picked this example specially because I had blending issues in the past so if you have blending issues you will know how to approach them.

This is my stitched Foreground and star circle images notice the two red arrows are places where the blending did not work so well

The first thing to try since the layers are aligned is the auto blend layers under the Edit menu.  Select the top and the bottom then navigate up to “Edit” and click under “auto blend layers”

 

This is the first tool I reach for when blending is the auto blend layers. Once the layers are aligned (and HDR-processed) thanks to PTgui the blending in Photoshop is usually easy.

If the auto blend function does not work it is time for some good old fashion hand blending. I will open the individual blend layers and the PTgui blended photo with blending issues.  I will use the photo with blending issues as the background and layer over the top the individual planes to blend by using a mask to gradually make the layers more transparent using a big soft brush.  I slowly make the seam fade or use the surrounding colors to add detail.  I will also use the clone stamp to replicate areas like clouds and blend them into each other.  This takes a lot of patience and practice to make some areas look “normal” but keep zooming in and out to see what affect you are having on the whole photo and local areas.

Adding the Star Circle

Once the blending the top and bottom image is finished the star circle can be added.  Since you exported it as a separate layer out of PTgui this can be brought into photoshop as a layer in the document.  Opening the photo as a layer then by changing its blend mode to lighten or screen blend mode then bright stars will out shine the dark background.  Thus adding the star circle to the finished photo was easy.  Don’t forget to mask out some of the areas were affected by the screen blend.  Say in cases of light pollution the foreground might be brighter then they should be.  Crop then your done.

Final image

This is the final image after the stars are added as a background layer and blended in using the screen blend mode.

Thanks for reading, as always comments and questions are encouraged.  If you have found this interesting please forward to your friends and follow us on Facebook. If you are interested in this topic (panoramas), night photography, shot planning, or super cool post processing techniques come and join us for a workshop.

 

Expanse of Waters – Night Vertorama

An example of using a panorama technique at night at Red Rock Canyon State Park, California. Coming soon a lot more information about night panoramas!

I created a Vertorama (vertical panorama) of the very last few shots before twilight began to overwhelm the stars. This is 3 separate images with about 4/5 overlap between each, but only the top image contains the star trail. You can see almost 3 hours worth of star trail taken here in this image.

I like the Vertorama technique here because it allows me to get the sky and the foreground in better context.

People often assume that the foreground was shot during the daylight, in fact the shortest exposure was 30 seconds and that was still at ISO 500! The very nearest components (the rocks and scrub) were taken during nautical twilight.

There was a tiny bit of moon (slender crescent) as well.