1. Reminder: Please use our affiliate links for holiday shopping!

Dark Frame Subtraction and bonus E-M1 vs E-P5 question

Discussion in 'Astrophotography' started by FlyPenFly, Mar 26, 2014.

  1. FlyPenFly

    FlyPenFly Mu-43 Veteran

    448
    Feb 15, 2011
    I'm going to be in a completely dark area next month and I plan to try out some dark frame subtraction but I'm a little lost on what the ideal settings would be. I plan to use a Oly 12mm F2 or Oly 12-40mm F2.8 at 12mm. Or maybe the Olympus 17mm F1.8.

    I also own an E-P5 and an E-M1. Which would be best?

    What exposure settings and how many of each type of frame would be recommended?
     
  2. aloysius

    aloysius Mu-43 Regular

    57
    Dec 18, 2012
    Nevada City, California
    My real name is unreal
    Sorry for the length of this post, but there is a lot to know to get the best astro images.

    The Noise Filter setting reduces noise in images by blurring them on a pixel scale. The Noise Reduction setting reduces noise by taking and subtracting a dark frame (shutter closed during exposure) from the light frame (shutter open). Because Noise Reduction exposes a dark frame in addition to the light frame, it takes twice as long. Only the subtracted image is saved to the memory card whether shooting JPG or RAW - the original light and dark frames are discarded.

    In astrophotography, dark frames are taken independent of light frames, and the dark frames are subtracted in post-processing. The big advantage is that multiple dark frames can be statistically averaged, for better overall noise reduction than subtracting single dark frames.

    For best results, leave Noise Reduction and Noise Filter turned off, and handle noise in post-processing. Take dark frames for the same shutter times as light frames, but with the lens cap on. The main requirement is that dark frames be taken at the same temperature as light frames, because noise approximately doubles with every 6 C increase of temperature. This is harder to do with a dSLR or MILC than a cooled astrocam because we don't have an easy way to see or control sensor temperature.

    At least one app for astrophotography will correct for temperature differences between light and dark frames: PixInsight. This requires taking a third kind of shot, called a bias frame. Bias frames are easy to take: set the camera at its fastest shutter speed and take a dozen shots or so, which will be averaged and subtracted from the light and dark frames in post-processing. This leaves only the temperature-dependent part of signal noise, which can be adjusted according to the sensor temperature.

    We don't usually see sensor temperature in EXIF data displayed by, e.g., Lightroom and Photoshop, but it is there and can be read by apps like EXIFTool. PixInsight reads this temperature value and adjusts to match the light and dark frames.

    I don't have an E-P5 but I do own the E-M5 and E-M1. Before the info that the E-M1 uses a Panasonic sensor instead of the Sony sensor of the E-M5 became public, I saw the difference in long astro exposures of 1 to 5 minutes duration. The Panny sensor is noisier in this type of photography.

    For that reason, I bought an E-PL5, which uses the same sensor as the E-M5 that I like for astrophotography, and sent it off to Life Pixel for the full-spectrum mod, for greater sensitivity to H-alpha from emission nebulae. You don't have to go this far, but my point is that the Sony sensor is better for long exposure astrophotography than the Panny sensor. I believe the E-P5 uses the same Sony sensor as the E-M5. Therefore, I would recommend the E-P5 for AP instead of the E-M1.

    With good technique using light and dark frames, the E-M1 is still fine for AP (and superb for normal high-ISO photography), but the E-M5, E-PL5 and (assuming it uses the Sony sensor) the E-P5 will be a bit easier to correct for noise.

    Optimum exposure settings depend on your target. Test shots are important. If you include a scenic foreground, expose to get that right. If you are shooting only the sky, expose to get the sky right. Expose for long enough to get some space between the histogram peak and the left (dark side) axis, for best noise reduction in pp. Watch what the few pixels at the bright end are doing. To preserve star colors, avoid overexposing them.

    Most often, a sky image will have too great dynamic range and either show more noise in the dark sky or blow out star colors. Then the photographer must decide which end of the brightness range to preserve. HDR techniques can be used to extend the dynamic range.
     
    • Like Like x 3
  3. aloysius

    aloysius Mu-43 Regular

    57
    Dec 18, 2012
    Nevada City, California
    My real name is unreal
    I would add that in-camera dark frame subtraction ("Noise Reduction") is better for astrophotography than no dark frame subtraction. If you don't want to do the detailed post-processing, you will still get good results by turning on Noise Reduction and accepting the doubled exposure times.
     
  4. FlyPenFly

    FlyPenFly Mu-43 Veteran

    448
    Feb 15, 2011
    Thanks, I did quite a bit more reading since I posted.

    I'm almost leaning towards selling my E-M1 and picking up a D600 or maybe a 6D to do a single exposure photo.

    The reason for this is that my primary objective is to get a creative foreground into the image with a starry night shot. I essentially want to recreate Starry Night in camera without severe post processing of masking and composting.

    Almost every photo I've seen of composted images that use a foreground and background do not look natural to me even if that's the optimal way to achieve that.

    It seems for that kind of single shot exposure, most people are using a full frame camera at around ISO3200 or ISO6400 and either the Rokinon 14mm F2.8 or 24mm F2 lens wide open and right around 30 seconds or less.

    I've been debating just renting a 6D and Rokinon wide angle to achieve those kinds of shots or perhaps selling my E-M1 and just keeping my E-P5 for MFT and getting a FF camera for more flexibility in my equipment capabilities.
     
  5. FlyPenFly

    FlyPenFly Mu-43 Veteran

    448
    Feb 15, 2011
    I don't think an Olympus E-P5 with a 12mm F2 or 17mm F1.8 could accomplish it with an acceptable level of noise? I could always just rent a 6D for the week and a Sigma 18mm F1.8 or Rokinon 24mm F2.

    According to DXO, the E-P5 suffers about 1 stop in SNR ratios at around ISO3200 and ISO6400. However, Dynamic range is much worse at those ISOs and it goes from 1.5 to 2 stops.

    I think in a squeeze, it might be possible judging from some of the exposure data that I could use the E-P5 with a F2 or F1.8 lens wide open to replicate those kinds of shots as linked above. Perhaps I could even match with the Voigtlander 17.5mm F0.95 lens if it was useable wide open to capture this kind of photo. It's theoretical though as I can't test it out where I live.

    950328.

    950329.
     
  6. aks6674

    aks6674 Mu-43 Veteran

    219
    Jan 14, 2014
    I tried a few of these a couple of weeks ago in the deserts of southern New Mexico and Texas (some of the shots are posted here - search for Big Bend, Carlsbad, and Milky Way if interested). One of the things I wish I'd done prior to the trip is to find a reasonably dark place to practice illuminating the foreground with a flashlight. I tried several different means of doing this, and none turned out as I wanted - most resulted in me blowing out whatever happened to be there.

    I think painting like this is a bit of an art, and may even require some diffusion for the flashlight if you want to give the landscape that 'moonlit' look. Fumbling around in the dark, trying to get it right before the clouds roll in or in your time window between the milky way rise and the sunrise is no time to learn - found this out to my detriment.

    Anyway, you may already have this covered but it's one of the things I really wish I'd done prior to my trip.
     
  7. aks6674

    aks6674 Mu-43 Veteran

    219
    Jan 14, 2014
    Forgot to mention, I did use the EM-1 with 12mm 2.0 on my trip, with noise reduction turned on. Might help you decide whether the performance is acceptable for you. I also tried the Rokinon fisheye, but was less pleased with the results.


    Sent from my Nexus 7 using Tapatalk
     
  8. bikerhiker

    bikerhiker Mu-43 All-Pro

    Dec 24, 2013
    Canada
    David
    I have reading and playing with this for quite awhile now since 2010 with my E-PL1 and have gotten pretty good results. You want to take the milky way I presume?!? First of all, even an old E-PL1, Olympus E-3 and E-5 can do the dark skies with some considerations to be successful.

    Cold night really helps maintain sensor temperature. If it's a warm night, start with a cold camera and when it gets warm, switch to an alternate camera. That's why I use both my E-PL1 and my Coolpix A for this purpose. Secondly, go for the FASTEST, BRIGHTEST and SHARPEST lens you've got in your arsenal. This way, you can use lower ISOs to reduce noise shooting in the dark, plus you DO NOT NEED to overuse the sharpening tool to increase acuity of your overall image. Using any sharpening tool can and will increase noise in your image. Thirdly, use an old school method like hyperfocal distance photography to increase depth of field to infinity while shooting wide open. For example, my pre-ordered Leica Summilux 15mm f/1.7 when shot @ f/1.7 will require me to pre-focus at the subject 30 feet from the camera's sensor plane, and this has the effect of extending the depth of field from 15ft to infinity. So you need to teach yourself how to set your lens to 30ft in the dark when you have nothing to focus at because my old E-PL1 does not have a distance scale. My Nikon Coolpix A does which makes hyperfocal distance photography work easy as pie! Turn off ALL noise reduction function, set color to neutral and gradient to normal and then shoot RAW. I found that shooting with the 14-42 kit lens @ 14 f/3.5 yielded ok results and is definitely not as good as my Coolpix A's 28mm @ f/2.8 which has a Nikon D7000 16MP sensor as well as no AA filter and Nikkor sharpest 28mm DX lens. However, I also found that shooting with my prime 25mm Pana Leica lens for a test yield impressively, sharp, constrasty results at about 1 stop slower than the Coolpix A. And then, I need to use DXO Optics Pro PRIME noise reduction technology to suppress the noise and the results were impressive enough that made me pre-order the Leica Summilux 15mm f/1.7 because of the experience I had with the 25mm version. Since 15mm is supposedly possess the same sharpness of the Leica 42.5mm Nocticron lens, I intend to use this advantage to start with the sharpest image I can obtain wide open and then the blur induced by DXO PRIME technology will still yield impressive results. I don't mind stars looking slightly soft. Some people went to great lengths to make them soft using the Kenko Pro-1D Softon filter. I just use DXO Optics Pro PRIME which softens the stars slighty as well as reduce noise.

    My Coolpix A with the APS-C sensor has about 1-1.5 stops advantage over my E-PL1 with the pre-ordered Leica Summilux 15mm f/1.7 and with that, I can easily retire my Coolpix A and only carry it if I need a double back up or an alternate swap camera when my E-PL1 heats up. With your E-M5, you should have no issues whatsover getting good star photos since the E-M5 has 2 stops better performance than my E-PL1. Again, people been doing this since 2010 with their older E series 43 cameras with some care. :)
     
    • Like Like x 1