1. Reminder: Please user our affiliate links to get to your favorite stores for holiday shopping!

Well that's an interesting effect of stacking on my E-M5

Discussion in 'Olympus Cameras' started by KBeezie, Sep 5, 2016.

  1. KBeezie

    KBeezie Mu-43 Top Veteran

    694
    Sep 15, 2012
    Grand Rapids, Mi
    Karl Blessing
    LQ7iSDZ.

    Sort of like a mosaic webbing in the shape of how light is distorted by the lens, not seen in a single image, but rather when stacked.

    Olympus 12mm f/2.0 @ f/2
    1/4 second
    ISO 12,800 (Noise Reduction and filter turned off in camera)
    Shot in Raw, Converted to TIF via ACR to be compatible with StarStax
    4 Dark Frames (capped shots)
    243 light frames

    StarStax (for stacking in lighten mode) + Photoshop (for color and resizing)

    A single raw file processed in ACR with the dark frame placed on top with subtraction blending for comparison.

    9SFUqkT.
     
    • Like Like x 1
  2. barry13

    barry13 Super Moderator; Photon Wrangler

    Mar 7, 2014
    Southern California
    Barry
    I guess those are diffraction patterns... Weird.
     
  3. Giiba

    Giiba Something to someone somewhere

    270
    Aug 19, 2016
    Burnaby, BC
    I don't lnow what to say about that pattern....

    But I do know for star pictures you're doing it wrong. With that combo you should be shooting at a way lower iso and a longer exposure time. I had great results shooting single exposures with the 12/2 at iso1600 and 10-13sec. For stacking try iso 1600 and 4sec or so.

    I suspect what is happening is some sort of high iso patterning being amplified by the stacking.
     
    • Agree Agree x 1
  4. Giiba

    Giiba Something to someone somewhere

    270
    Aug 19, 2016
    Burnaby, BC
    I have never tried stacking, but here is an example from my e-m1 and 12/2.

    iso1600, 20s
    P3280163_DxO_4.

    Note the forum downsizing is very unkind to this sort of shot. I have this printed at 24x18" and it is gorgeous.
     
    • Like Like x 3
    • Agree Agree x 1
  5. AllanG

    AllanG Mu-43 Regular

    30
    Aug 26, 2014
    Brisbane, Australia
    Allan
    It's not necessary to go with such high ISO and low shutter speed. It's better to use a moderate ISO (1600-3200), 15-30 sec exposure with in camera noise reduction for the EM5. This will give a good single image which can be processed in PS, or if you want you can stack at least 10-60 images to pull out good star or galaxy colour or even to turn them into an AVI or star trails.
    Do you have a UV filter on the 12mm? - this can sometimes give an interference pattern.
     
    • Agree Agree x 1
  6. KBeezie

    KBeezie Mu-43 Top Veteran

    694
    Sep 15, 2012
    Grand Rapids, Mi
    Karl Blessing
    Thanks for the unsolicited critique for my 'experiment' of what it would have looked like at really high iso.
     
    • Funny Funny x 1
  7. KBeezie

    KBeezie Mu-43 Top Veteran

    694
    Sep 15, 2012
    Grand Rapids, Mi
    Karl Blessing
    You should try stacking, that's a lot of noise/patterns esp in the dark areas.

    Downside to in camera noise reduction is the duration of capture is doubled, when you can easily do the same thing taking 5 or 6 dark frames before or after the shooting at the same exposure. Rather than rely on the in-camera dark frame subtraction for each image.

    Also your exposure won't work here, too high a light pollution, hence the experiment of 'what if'. Wasn't trying to do it "correctly" to begin with.
     
  8. KBeezie

    KBeezie Mu-43 Top Veteran

    694
    Sep 15, 2012
    Grand Rapids, Mi
    Karl Blessing
    I'm aware, that wasn't the purpose of this experiment, I was checking to see how noise reduction worked with stacking, though in photoshop it would just be doing a median/mean stacking with a smart object (of all the images loaded into layers then converted). It does work, but I wanted to see how well it worked against ISO 12,800 with something that could be considered noise by the software (ie: specks of stars). But course photoshop couldn't correctly auto-align the layers because it kept complaining about 40% overlap despite being like 98% overlap.
    DSS could align them though, but you don't want foreground objects in that. So instead I used StarStax just to see what it would look like stacked given some dark frames to process (in photoshop you can just do dark frames on top of a single long exposure using subtraction blending for similar result).

    Sort of, there's a Sigma "Protector" filter on front that cost me like 45 back when I got the lens. So that's likely what's going on.
     
  9. Giiba

    Giiba Something to someone somewhere

    270
    Aug 19, 2016
    Burnaby, BC
    Fair enough, you didn't mention you were experimenting. I was trying to suggest a way to avoid the weird patterning you found. It is just that iso that high isn't helping whatever the case. Though I think AllanG nailed it with the filter thought.
    See: www.Clarkvision.com: ISO and Digital Cameras, ISO Myths

    You'll note I suggested 4sec or so, which would need experimentation to find the max exposure time for your sky conditions (ie light pollution). My pic was taken well away from a light pollution source so I could use a much longer exposure, and I shoot with an e-m1 which is known for it's horrendously bad long exposure noise so I'm stuck with the in camera noise reduction for decent results.
     
  10. PhotoDmiK

    PhotoDmiK Mu-43 Regular

    83
    May 26, 2015
    Do you have IBIS turned on?
    As for pattern it looks like distortion correction.
     
  11. KBeezie

    KBeezie Mu-43 Top Veteran

    694
    Sep 15, 2012
    Grand Rapids, Mi
    Karl Blessing
    No, I avoid turning it on if I'm mounting on a tripod just because it can counterproductive.
     
  12. KBeezie

    KBeezie Mu-43 Top Veteran

    694
    Sep 15, 2012
    Grand Rapids, Mi
    Karl Blessing

    Try dark frame subtraction, much faster and may product as good or better a result if you just take 5+ shots with the lens cap on (All other exposure the same), layer them in photoshop above the image with subtraction blending.
     
  13. PhotoDmiK

    PhotoDmiK Mu-43 Regular

    83
    May 26, 2015
    You take set of photos with very little sensor movement. Could be vibration caused by shutter. Noise multiplied by stacking shifted images and you see it modified after distortion correction.
     
  14. KBeezie

    KBeezie Mu-43 Top Veteran

    694
    Sep 15, 2012
    Grand Rapids, Mi
    Karl Blessing
    I used an intervalometer and a 1/2 second shutter lock.
     
  15. alex g

    alex g Mu-43 Top Veteran

    538
    Mar 30, 2016
    New York / Bath
    It's the entrance to a worm-hole, obviously.
     
    • Funny Funny x 2
    • Winner Winner x 1
  16. KBeezie

    KBeezie Mu-43 Top Veteran

    694
    Sep 15, 2012
    Grand Rapids, Mi
    Karl Blessing
    Oooooh, good thing it wasn't a little lower to me.
     
    • Agree Agree x 1
  17. eteless

    eteless Mu-43 All-Pro

    Jun 20, 2014
    My first thought was an interference effect caused by diffraction, however I quickly discounted that due to it's size. The sensor has an active area of 17.3 by 13mm which would mean the airy pattern spacing was 0.14 to 0.18mm, the wavelength of light likely needed to cause an interference pattern this large would therefor be in the deep infrared range (50-75µm) where a silicone sensor has no sensitivity (never mind glass being opaque to it).

    My second thought was that it was an internal reflection in the lens of some sort so I pulled up the design and realized the rear most elements are both positive. The edges of the lenses are a greater distance apart than the center of the lenses, so while the center retains the original orientation (it's close enough to not invert) the outer edges flip upside down, this causes the circle blob (where the circle meets the outer corners is the transition between the two).

    To prove my hypothesis I mounted a large plano convex lens (I think it's a +10 diopter, it's a junk lens - a magnifying glass would likely work just as well if someone wants to test this themselves) around 10cm in front of a monitor on which I displayed a grid of lines, I then took a picture from around two meters distance with a Sigma 300mm stopped down to f/32 to gain enough depth of field (the planes of focus between the screen and focal point of the lens are radically different).
    positiveelement.
    As you can see the resulting pattern was reasonably close to the image provided above.

    I'm not completely certain the exact source however it seems likely that it's the micro lens array in front of the sensor which bends light to enter at a better angle. These lenses would cause a specular reflection of any stray light hitting them, in normal operation this tiny amount of light wouldn't cause any issues as the antireflection coatings on the lens would lower it by around 99% (the light needs to bounce off something). As this is a stack of 243 frames the light is amplified and can be easily seen.

    I have seen a related effect caused by the Sun being in frame (causing the specular reflection off the micro lenses to be very bright) on other cameras, off the top of my head this includes the Sony RX10 and Olympus EPM-2 (only with the 14-42R kit lens heavily stopped down, wide open it doesn't happen). It probably happens with all cameras with a micro lens array to some degree however it's so rare to push them to a large enough extreme that it would become visible.

    TL:DR: it's probably a reflection of the micro lens array on the sensor.

    ...and now my lunch break is over and back to work... ...should really have eaten lunch instead of typing this up...
     
    • Like Like x 3
    • Informative Informative x 2
  18. KBeezie

    KBeezie Mu-43 Top Veteran

    694
    Sep 15, 2012
    Grand Rapids, Mi
    Karl Blessing
    Hrm, in such a case if I were to try to re-produce it, I should try a couple different lens, and if the pattern stays the same with the exaggerated sensitives then we can probably conclude that is what is happening.

    It's interesting to see, and it also re-iterates the suggestions against such high sensitivities for such a purpose.
     
  19. eteless

    eteless Mu-43 All-Pro

    Jun 20, 2014
    I pulled the designs for m4/3 lenses made in the last 5ish years and there isn't that many which have a lens arrangement which could cause this kind of effect, they include:
    Olympus 14-42mm F3.5-5.6 (100% certainty to cause this and other problems however only the very first discontinued version, all three later versions were redesigned)
    Olympus 8mm f1.8 (due to the diameter of the lens elements I suspect it's very unlikely to occur)
    Panasonic 8mm f3.5 (same as the Olympus 8mm)
    Olympus 12mm f2 (before today I would have said no, however I'm going to say maybe)
    Voigtländer 10.5mm f0.95 (In normal usage it has nothing at all to suggest it would however afaik no one has stacked hundreds of frames to date)
    Voigtländer 17.5mm f0.95 (basically the same as the 10.5, In normal usage there's no hint at all)


    I think most other lenses should be immune based on their designs however I must stress that even if these lenses do share this trait it's going to take the same mistreatment using several hundred severely underexposed frames combined for anything to show up (if anything at all), this is really not normal treatment for a lens and even if a lens were prone given this treatment I would be perfectly happy to use it in every day situations (with the exception of the original 14-42mm, it really is/was a terrible design).
     
  20. Giiba

    Giiba Something to someone somewhere

    270
    Aug 19, 2016
    Burnaby, BC
    I'm very curious to learn more about this design you mention, I've looked at the lenses you mention but I might as well be reading any number of languages I do not speak. Your magnifying glass experiment certainly seems to be on to something.

    Could you maybe post a pic of what you speak of? I just don't see what sets those particular lenses apart from other m4/3 lenses.