Stars over Wellington

Discussion in 'Astrophotography' started by wjiang, Mar 14, 2015.

  1. wjiang

    wjiang Mu-43 Hall of Famer

    I took these on my recent holiday in windy Wellington, New Zealand. Despite remaining inside civilisation, these were the darkest skies that I have come across so far, much better than what I get in Christchurch. So, never mind cold, wind, and clouds, I ventured out into the night with my trusty Slik travel tripod and E-M5...

    The following were taken from the roof of The Keep (, where I stayed for a night. It's a great place by the sea, over the hill from Wellington city proper - thoroughly recommended.

    7.5mm FE, 4x 25s stack at f/3.5, ISO1600:

    And this is what f/1.7 versus f/3.5 reveals...

    PL 15mm, 8x 10s stack at f/1.7, ISO1600:

    Another night: The Krupp Gun, overlooking Wellington from the Botanical Gardens. Noticeably more light pollution from here.

    7.5mm FE, 6x 10s stack at f/3.5, ISO1600:

    Finally, on a different night again, this is the old fort on Wrights Hill (with a modern microwave communications tower). Less light pollution here, but a major case of cloud meant I had to time shots quite deliberately. The summit is sufficiently high that I was in the cloud a lot of the time. Getting enough frames for a stack proved troublesome!

    7.5mm FE, 3x 13s stacks at f/3.5, ISO800:
    Last edited: Mar 14, 2015
    • Like Like x 14
  2. Speedliner

    Speedliner Mu-43 All-Pro

    Mar 2, 2015
    Southern NJ, USA
    Real Name:
    Im lucky to see 10 stars where I am. Great pics!!!
  3. maritan

    maritan Mu-43 Veteran

    Oct 30, 2014
    Great shots! I love this stacking technique. I'm hoping to get some practice soon.

    Thanks for sharing.
  4. wjiang

    wjiang Mu-43 Hall of Famer

    I forgot to mention that I recently learned the Photoshop command to load as stack (with auto align) which makes merging the foreground stack and background stack a bit simpler.

    I'm really happy with how this technique is working out, even on our supposedly tiny m4/3 sensors with poor high ISO performance ;-)

    Unfortunately it's kind of making me have tinges of GAS for the upcoming 8mm f/1.8 fish-eye and 7-14mm f/2.8...
    • Like Like x 1
    • Useful Useful x 1
  5. maritan

    maritan Mu-43 Veteran

    Oct 30, 2014
    I'll have to check out how to do that. I have CS5, so I can probably do something similar to what you have. Do you have a link that you might be willing to share?

    And, yes, I have serious GAS for the 7-14mm f2.8 too - the fish eye not so much yet.
  6. wjiang

    wjiang Mu-43 Hall of Famer

    I don't normally bother converting to smart object if I'm manually blending two stacks from DSS.
    • Appreciate Appreciate x 1
  7. walter_j

    walter_j Mu-43 Veteran

    Sep 10, 2013
    Hagwilget, B.C., Canada
    Real Name:
    Have you ever tried making a panorama with astrophotography? Trying to rationalize not buying the Kowa 8.5. I also don't understand the need for a zoom UWA, which undoubtedly makes it slower and more expensive.
  8. wjiang

    wjiang Mu-43 Hall of Famer

    Haven't tried, but with so many long exposures there's already more than enough movement to require a separate star and foreground alignment. It would also take a very long time and result in way more data to process. The other thing I noticed is that even a portrait orientation panorama sweep with a 12mm doesn't cover as much view as the fisheye. Interesting idea though, I might give it a go with my PL15 to see how it compares.

    I actually think a UWA zoom makes a lot of sense for general purpose use. Generally with UWA and supertelephoto you have no ability to frame by moving (at least with supertelephoto you can crop), so having the ability to zoom is really useful.
    • Like Like x 1
    • Agree Agree x 1
  9. Edmunds

    Edmunds Mu-43 Regular

    Oct 16, 2012
  10. wjiang

    wjiang Mu-43 Hall of Famer

    I first stack with auto-align by stars using Deep Sky Stacker (free). This gives sharp stars but if there's foreground, that ends up trailing. If there is foreground to worry about, I do another stack without alignment to get foreground sharp (but stars trailing), then blend those two in Photoshop manually. Hope that clarifies things.
    • Like Like x 2
  11. maritan

    maritan Mu-43 Veteran

    Oct 30, 2014
    I have a stack of 20 pictures that I'd love to be able to stack using DSS, but no luck. I have to run DSS via Wineskin on my Mac (if that matters?), and I'm probably missing some step(s?) or settings, so I'm just not getting meaningful outputs yet.

    Maybe the problem is that I only have light frames (no darks, flat bias etc.)? Have to re-read the manual.
  12. wjiang

    wjiang Mu-43 Hall of Famer

    It should still work even with only light frames. That's how I prefer to do things with a fisheye actually, because DSS seems to have more diffulty coping with dark frame subtraction with a fish eye (it wasn't really designed for super wide angle stacking). I get the camera to do its own dark frame subtraction per frame and stack the output as light frames, this way the dark frame is perfectly matched with each light frame.

    One thing I have found is that if you underexpose the light frames, DSS can a) have difficulty finding stars and b) result in unbearably noisy stack output if you attempt to lift exposure afterwards. In these cases, I have found that if you lift exposure in Adobe Camera Raw and convert to DNG or 16-bit TIFF, DSS will work a lot better when you feed it the lifted output.

    Secondly, if there are significant numbers of distant city lights in the frame, DSS may get confused and try to align those instead of stars. In these instances, I use the ACR to crop out city lights before stacking. They are still there in the unaligned ground stack that gets blended back in.

    What looks wrong about it? The 32-bit stack output will need significant post processing on it - as it is the output will look extremely flat and low contrast since it's colour balanced for the sky background and has a crazy 32-bit gamut that nothing can display properly...
  13. maritan

    maritan Mu-43 Veteran

    Oct 30, 2014
    @wjiang@wjiang - First time I tried, I got a flat gray output. Nothing in it except a gray screen.

    For the second attempt, to save some time, I used just two of the 20 pictures and attempted to stack that and I got an error message saying it would use only one of them to stack...

    I'm definitely missing a few things, but I'm not sure what yet.
  14. wjiang

    wjiang Mu-43 Hall of Famer

    Very strange... never had either happen before.
  15. maritan

    maritan Mu-43 Veteran

    Oct 30, 2014
    So, I tried it again, on a Windows PC. Here's the result I got. Horrible artifacts and the bottom is cut off. But, most likely because the overexposed moon is confusing the program? I'll have to try again when the new moon comes around (can't go try with tonight's new moon - too cloudy. Bummer.)

    Attached Files:

  16. wjiang

    wjiang Mu-43 Hall of Famer

    Oh, having the moon in the shot could be a problem. How many stars did DSS find? It should be getting hundreds at least for it to be effective - I can't see many at all in the JPEG. The moon is way brighter than the background stars, you'll have to completely over expose it in order to get the background stars to be bright enough. The moon light will also act as significant light pollution on the atmosphere, drowning out the dimmer stars. As an aside, I also see flare spots caused by the super bright moon.

    Where was the moon supposed to be in the frame? Did you try cropping it out?
    Last edited: Mar 21, 2015
  17. maritan

    maritan Mu-43 Veteran

    Oct 30, 2014
    DSS found 470 stars! I thought that would be enough and more. And yes, the moon did cause flare with my 12-40 pro. The moon was right next to the Milky Way.

    The final result from DSS looks a LOT worse than just one of the pictures processed (which also admittedly is not the greatest). I'm still figuring out how I want to process Milky Way pictures, but here's one attempt at it.

    • Like Like x 4
  18. wjiang

    wjiang Mu-43 Hall of Famer

    Wow, that's excellent, DSS should have no trouble with exposures like that!
    Do note that the direct output from DSS is completely useless without further PP. Your example earlier just looks to be an alignment problem, the rest is pretty much what I would expect from a direct DSS output. It's a 32-bit floating point TIFF, so even though it looks super dark and flat it shouldn't be a problem. If you don't mind uploading 2 of the RAW frames, I could take a look at where it might be going wrong.
    • Appreciate Appreciate x 1
  19. maritan

    maritan Mu-43 Veteran

    Oct 30, 2014
    Sent you a "conversation." :)
  20. wjiang

    wjiang Mu-43 Hall of Famer

    I'm posting the findings regarding maritan's RAWs with permission, in case it helps others...

    Nothing was really wrong with the RAW files, other than a little more noise than I'm use to with my E-M5. I suspect that's what's confusing DSS, as it fools it into thinking those noise spots are stars to falsely align on. Here's what a test run through with 3 RAW frames was able to yield:



    I just loaded the RAWs in. Generally there is no need to make many adjustments before stacking - it's best to let DSS itself do as much as possible, except if the frames are way under exposed (lift exposure, but don't clip stars), or if there is too much noise (apply very light NR).

    Register Settings:
    • Automatic detection of hot pixels = ON
    • Star detection threshold = 20%
    • Reduce noise by using a median filter = ON
    With those settings, it only found about 100-200 stars per frame - sounds about right, not confused by false stars.

    Stacking Parameters - all set to recommended settings, except:
    • Alignment = Automatic for stack #1 (sky)
    • Alignment = No Alignment for stack #2 (ground)
    After stacking, I don't bother adjusting further in DSS. I simply save the default output, choosing to embed adjustments in the saved image and not apply them. The stacks will be blended in Photoshop so it's important both the ground stack and sky stack have no adjustments (and therefore will look like they're supposed to be from the same images).

    1. Scripts->Load Files into Stack (Attempt to Automatically Align Source Images).
    2. Crop the non-overlapping edges off.
    3. Use a layer mask to blend in the ground and sky layers.
    4. Flatten image - we now have a flat looking 32-bit file with more adjustment latitude than RAW. It will probably overwhelm Photoshop's sliders.
    5. At this point I tend to convert to 16-bits per channel. This will bring up Camera RAW, which is nice because we can use it to bring the 32-bit file into normal editing range after the conversion. No point trying to get it perfect at this stage. I tend to just fix the WB, vignetting, and in general even out the histogram as much as possible, it doesn't matter if the sliders are maxed out and it still isn't enough.
    6. Once in 16-bit mode, normally editing tools and techniques apply. Hopefully the histogram is now even enough such that the normal slider adjustment ranges are enough. The image will probably need NR and sharpening at this stage, as DSS will not have done any of that.
    Turned out the difference was due to not having the latest version of DSS (3.3.4)!
    Last edited: Mar 31, 2015
    • Like Like x 3