12
\$\begingroup\$

Looking at the winner of Astronomy Photographer of the Year, I'm wondering, how can one produce shots like these:

http://www.rmg.co.uk/whats-on/exhibitions/astronomy-photographer-of-the-year/winners-2011/special-prizes/

Specifically I'm referring to the "People and Space: winner" image. When the winner say that he used "525 separate exposures" to create the image, what does this mean? Any tutorials on this workflow?

I suppose this image uses the same method:

http://www.universetoday.com/86472/are-you-the-next-astronomy-photographer-of-the-year/

I mean, even if it is really dark, how could one possibly get that much detail from just one exposure without stars beginning to trail?

\$\endgroup\$
8
  • \$\begingroup\$ As far as I can tell it's a composite photo, with the stars being added on later, but as I don't know the techniques I won't put it in an answer \$\endgroup\$
    – Dreamager
    Sep 21, 2011 at 16:50
  • \$\begingroup\$ True, that is probably one way. Even so, let's focus on the "525 separate exposures" part and/or creating the effect from scratch. \$\endgroup\$
    – Figaro
    Sep 21, 2011 at 17:07
  • \$\begingroup\$ My guess is that he averaged all the exposures after aligning them so that the starts don't trail. The foreground is probably a separate exposure. \$\endgroup\$ Sep 21, 2011 at 17:08
  • \$\begingroup\$ On topic of avoiding star trails, see How can I avoid star trails without an expensive tracking mount? \$\endgroup\$
    – Imre
    Sep 21, 2011 at 18:09
  • 1
    \$\begingroup\$ The "525 separate exposures" appears to be a error on the part of the website - the photographer says on flickr it was a single exposure! \$\endgroup\$
    – Matt Grum
    Sep 22, 2011 at 7:24

6 Answers 6

7
\$\begingroup\$

A little bit of research goes along way...

The image with the figure in the lower right that won the "people and space" category was not created from 525 separate exposures as the article claims, but one single relatively short exposure. From the photographer himself, via flickr:

The setup was pretty simple... I found the foreground hill where I could stand silhouetted against the night sky, I set the camera on 10 second self timer for a 30 second, f/2.8, ISO 6400 exposure, then I walked into the frame and stood still until I heard the shutter close.

I suspect the "525 images" got misplaced and belonged to one of the images from the "Robotic Scope" category. These are images of far away objects through proper telescopes with tracking mounts. Many exposures are needed due to the dimness of light reaching Earth from these distant structures.

\$\endgroup\$
6
  • \$\begingroup\$ I'm sorry but I don't believe him in the slightest. There's no way he got that clear a pic of the milkyway with a single 30s exposure :o \$\endgroup\$
    – Dreamager
    Sep 22, 2011 at 11:27
  • 1
    \$\begingroup\$ @Dreamager What motivation does he have to lie about the exposure time? Unless you're saying he comped in the milky way which is a serious allegation with no evidence. 30 seconds at f/2.8 is a lot of light when you amplify it sixty four times (in camera, possibly more amplification in post). \$\endgroup\$
    – Matt Grum
    Sep 22, 2011 at 12:20
  • \$\begingroup\$ Having a read around it does seem doable. I guess I didn't hold out much hope for the faintness of the milky way showing through the noise of ISO6400. SOmething for me to try now ;) \$\endgroup\$
    – Dreamager
    Sep 22, 2011 at 12:37
  • 1
    \$\begingroup\$ @Dreamager You're looking at a 0.3 megapixel image, downscaled from a 21 megapixel full frame DSLR image. That has a considerable impact at smoothing noise. You could bin 70 original pixels together to produce each output pixel at that resolution! \$\endgroup\$
    – Matt Grum
    Sep 22, 2011 at 12:49
  • 1
    \$\begingroup\$ @Dreamager: It doesn't take a lot to get a decent wide-field night sky photograph. With a good lens and camera, good milky way shots usually take 25-35 seconds. Whether you get star trails or not is a function that involves focal length, and the lens used was very wide angle. At 16mm on a FF sensor, its unlikely you would get much star trailing even after a 40-45 second exposure. \$\endgroup\$
    – jrista
    Sep 22, 2011 at 23:41
5
\$\begingroup\$

Well, its (the first image) definitely a composite photo (assuming he stacked as the article said...which it turns out he didn't and the article lied lol) - the foreground and the stars are not from the same set of exposures - so lets just focus on the star part. Use any method you want to put the foreground and background together. If you're stacking photos, and there's a clear, sharp landscape in it as well - its a composite.

The 525 separate exposures was mostly likely done either through mosaic's as cmason said, but for widefield astro like this, its also common to use a tracking system like AstroTrac and a stacking program like DeepSkyStacker. You'll take many, many shots of the exact same area, using a mount that rotates at the sidereal rate to keep the same view. Then, you stack them together - think kind of like stacking multiple transparencies together. This makes it easier to 'pull' and 'push' the curve or levels around without increasing noise too much.

Good information in this question as well.

\$\endgroup\$
3
  • \$\begingroup\$ The image from the second link in the question is indeed a single exposure, as its from the TimeScapes time-lapse night photography videos. The foreground is lit by an artificial lighting setup. \$\endgroup\$
    – jrista
    Sep 21, 2011 at 23:18
  • \$\begingroup\$ @jrista I should be more clear - I was referring to the questions about stacking (and the fact that it included a landscape) - the author assumed the second one was, but apparently wasn't pertinent to his real question. \$\endgroup\$
    – rfusca
    Sep 22, 2011 at 1:47
  • \$\begingroup\$ Well, @MattGrum found outthat the article had a misprint regarding the stacking of images. I'm going to leave this up, as its still pertinent if you wanted to stack but its just not how that image was done. \$\endgroup\$
    – rfusca
    Sep 22, 2011 at 13:18
3
\$\begingroup\$

Regarding the image from the second link, I'm pretty certain that is a single exposure. The photographer, Tom Lowe, is a renowned night and astrophotography photographer, and has a series of small video clips that will eventually be part of a production called "Timescapes". Tom uses rather elaborate lighting and camera rigging setups to capture his time lapse videos, and in the case of the photo you linked, I do believe there were a couple artificial lights illuminating the foreground tree. At that point, capturing the level of detail in the tree is a pretty strait forward matter, as the exposures are quite long to capture the night sky. There are also cases where foreground objects in his time lapse sequences are lit by the moon, which is often (initially) behind the camera near the opposite horizon. That provides quite a bit of illumination for a long exposure shot, and sometimes looks like there is artificial lighting. If you watch his sample videos on his site, the actuality of his scenes lighting becomes much more apparent as each sequence progresses.

Regarding the pictures from the first link, its harder to tell. Its not uncommon for astrophotographers to take two exposures, one tuned for the night sky, and one tuned for the foreground, whatever it may be, and manually blend the two with post-processing tools. Photoshop is a common tool when you are not doing long-exposure night sky photography. More elaborate tools, such as DeepSkyStacker, are often used when taking multiple exposures while tracking the sky with a tracking camera mount, to reduce noise and maximize saturation.

Regarding 525 exposures on that one photo, I don't see how that is reasonable. For one, a simple stack of 8 rather "short" exposures (short is a relative term here...you need at least a few seconds at a very fast aperture to capture a bare minimum of detail in the night sky) is usually more than enough to stack a nice, brilliant, colorful and saturated photo of a stellar object. If we assume the photographer used a tracking mount, then the landscape would not have remained stationary as the camera tracked the milky way across the sky. The landscape (as well as the silhouettes of people) is quite clear. With a tracking mount, then there really wouldn't have been much need for multiple exposures, certainly not 525 of them...you could do far fewer exposures for longer times (say, 60 seconds each), and stack them to get a much better result. Even a single exposure of say 5 minutes at ISO 100, on a tracking mount, would produce a better result. The details of the photo on flickr state that a 16-35mm wide angle lens (given the camera was a Canon 5D, probably the Canon EF 16-35mm L) was used. Barrel or pincushion distortion is apparent around the edges of the photo when the largest is viewed on Flickr, so its highly unlikely a telescope was used to take narrow FoV photos that were stitched into a mosaic. Its certainly possible, however most definitely not necessary, to stack 525 photos to get a shot like that.

\$\endgroup\$
10
  • \$\begingroup\$ Aye, yes...however I edited the last paragraph to be simpler anyway, and realized that as I was editing. \$\endgroup\$
    – jrista
    Sep 22, 2011 at 0:15
  • \$\begingroup\$ Its definitely on the high side, but certainly not unreasonable. I've seen members of my astro club take pictures all night of the same object. It helps most with pulling deep space objects out. Definitely more than a single exposure. \$\endgroup\$
    – rfusca
    Sep 22, 2011 at 1:50
  • \$\begingroup\$ I suspect it is taken on more that one nite \$\endgroup\$
    – cmason
    Sep 22, 2011 at 2:18
  • 1
    \$\begingroup\$ @jrista turns out you were right - they must have mixed up the captions, the "people and space image" wasn't 525 exposures at all, it was a single 30 second exposure! \$\endgroup\$
    – Matt Grum
    Sep 22, 2011 at 7:10
  • 1
    \$\begingroup\$ @Figaro: A strong contributing factor to the amount of detail was the camera body. The Canon 5D Mark II is a very high resolution sensor with some pretty stellar high-ISO performance. I have a Canon 450D, and for an identical exposure with the exact same lens (which I do own and have used for milky way photography), my camera would exhibit far more noise at lower ISO, with far less saturation. Having the right gear is really a bonus when it comes to photographing the sky. Its not really a necessity, but it is a bonus. \$\endgroup\$
    – jrista
    Sep 22, 2011 at 23:45
3
\$\begingroup\$

I've only made a few attempts at this with some success. I also used deep sky stacker. I used a standard ballhead tripod, a canon 7D, and a 17-55mm f/2.8.

There is a little more to it then just stacking the exposures taken (light frames). Steps taken: I took 20 exposures @ f/2.8, 15/sec, ISO 1600, 17mm (even this gave me a small amount of trail on my stars). IMMEDIATELY after I took 15 "dark frames". To do this you leave all of your settings the same, put the lens cap on, then take your exposures. The point of the dark frames is that it essentially takes a picture of the camera noise, hot pixels/dead pixels. Deep Sky Tracker will compile those into one dark frame to subtract that information from your set of light frames. It's important to do it right after your exposures while your sensor is still the same temperature, thus the noise level is the same.

When using your stationary tripod, you would not be able to take 500+ exposures, and here's why: the only option you have is for DSS to locate all the stars in your frame (even tells you how many) and align them. Ultimately you have less stars than in a single image because of how the stars drift. If there are stars that drif out of your frame, and new ones drifting in, it won't stack those and cuts them out of your final image. SO the fewer frames you can get away with, the better for that reason. What you'll notice in the final product is that the stars are stacked, but any forground will have moved instead. Thus the need to make a composite image.. one of the stacked stars, then adding in the foreground, usually by using layer masks in Photoshop.

From what I've learned in my attempts, I'm almost positive the first image was done with a wide angle lens using a tracking tripod. If it had been stationary, after taking his exposures the milky way would have moved all the way accross his frame, if not completely off of it and he would have been left with little to no stars DSS would recognize as being the same, and stackable.

SO it can be done with a standard tripod. Location is the key. Get away from any light polution ( I mean ALL the way away), check a moon phase calendar and plan it for a night that there will not be a moon in the sky, get as wide as you can, with the aperture as wide open as possible, and hike up your ISO as high as is practical. My next attempt will be ISO 3200, maybe 6400. There is some math you can use to judge how long you can expose for without creating star trails. I didn't use it. I simply took a test exposure then zoomed in on the preview, adjusting until I found an acceptable setting.

Give it a shot and have fun. :) oh, and don't be alarmed when you see red pixels in your display after shooting for a while. You can spend six grand on a Canon 1Ds MKIII and still have dead pixels. It's unavoidable. ;)

I'm no expert and I certainly haven't figured it out completely. I'm hoping someone will reply with some suggestions for me as well.

-Rocco

EDIT: be sure turn high ISO and long exposure noise reduction OFF. It's unecessary using this method.. and takes just as much time to apply it as it does to take the exposure. The less time in between exposures, the less star drift there will be.

Also worth noting.. DeepSkyStacker is a free program.

\$\endgroup\$
9
  • \$\begingroup\$ You theoretically could do it without tracking by using an ultra wide and stacking several different nights together - but it probably was on a tracker. Dark frames are indeed a good idea. Exposure time math is 600/focal length I believe - as a rough estimate. Opening your aperture wide open will make even the stars a bit soft, they won't appear to 'shine' - try to stop down once or twice. High iso is one way, but lower iso and more stacks works really well - less noise and more DR per shot. \$\endgroup\$
    – rfusca
    Sep 22, 2011 at 5:43
  • \$\begingroup\$ Rfusca.. couldn't reply from my phone.. annoying.. but thanks for the suggestion. I never would have guessed at that. I wonder how this would work: more light frames say.. 40-50, ISO 640 or 800, 15 sec, f/4 or so. I'm most likely going out again Friday night due to the moon not rising until around 4 am. I'm definitely going to give that a shot. I definitely know what you mean about the soft appearance, but I guess I figured that was pretty much unavoidable. I thought that the higher the ISO and the wider the aperture, the more stars your sensor picks up. I'm excited to try this out. Something \$\endgroup\$
    – Rocco
    Sep 22, 2011 at 6:10
  • \$\begingroup\$ well there's a certain truth to 'higher iso and wider aperture' for more stars, but it really makes sense for single shots. Once you start stacking you can set your ISO whereever you want depending on how long you're going to shoot for. If you've got all night, then setting it to ISO 100 and doing 50 shots at ISO 100, would get you more than if you've got a couple of hours at do 75 shots at ISO 800 - its all a compromise. In regards to focus, read here - astronomy.stackexchange.com/questions/180/… \$\endgroup\$
    – rfusca
    Sep 22, 2011 at 6:27
  • \$\begingroup\$ Makes sense. I guess I just need to find that balance. If it would be helpful to the original poster of this question, I'll try a few different configurations and post links to the results. Something I've often wondered: would you be able to take a single frame, duplicate it on your computer, and stack those? Or would that defeat the purpose? Do those programs work by negating the differences in each frame relative to the stars it recognizes? \$\endgroup\$
    – Rocco
    Sep 22, 2011 at 6:40
  • \$\begingroup\$ If you're being really precise you should interleave the dark and light frames, as the sensor temp for your first exposure will be lower than that of the 20th. \$\endgroup\$
    – Matt Grum
    Sep 22, 2011 at 6:57
2
\$\begingroup\$

I'm confused too by the claim that it's "525 separate exposures". It looks like just one to me. In a location like that (away from major light pollution) a single exposure at around 30s, f/2.8 and ISO 3200 will easily capture that much detail. This photo of the Milky Way on Flickr was taken at 30s, f/4 and ISO 6400 and the Milky Way is clearly visible.

Stacking multiple exposures (using software such as Photoshop or Startrails) is usually done to create light trails, where the stars forms streaks across the sky. This example is a stacked series of 126 frames.

I'm no expert though: if someone knows how a shot like that can be put together from 525 frames I'll be interested to hear about it.

\$\endgroup\$
1
  • 4
    \$\begingroup\$ Stacking is very common besides creating light trails. Notice in your example from flickr of the single shot, it doesn't scale well at all when viewed even reasonably large. There was a huge amount of NR applied and a star trails formed. Stacking multiple shots allows for much more detail with less noise and less trails. Stacking with landscape involves masking out some stuff near the landscape and typically using something like deepskystacker. \$\endgroup\$
    – rfusca
    Sep 21, 2011 at 21:53
2
\$\begingroup\$

Using mosaics, what we also call composites. In this case, the photographer most likely used the camera with a lens to capture the landscape, and also took shots with a telescope. The telescope has a tiny field of view, so in order to get a wide field 'whole sky' image like this, you would need to take many shots of each portion of the sky visible in the telescope, and then composite them all together in Photoshop to replicate the whole sky visible to the eye.

Read more here: http://www.robgendlerastropics.com/Article3.html

\$\endgroup\$
1
  • \$\begingroup\$ I'm not sure this level of compositing would be allowed within the rules of the competition, plus you'd still need to stack exposures to reduce noise if you're using a telescope so I'd bet the 525 exposures were with the camera on a tracking mount, using the same lens. \$\endgroup\$
    – Matt Grum
    Sep 22, 2011 at 6:43

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.