Tags:
  1. Mick West

    Mick West Administrator Staff Member

    When you take a photo or video of a bright light (or even just look at it) the light often appears a lot bigger than it actually is. The phenomenon is well known, but not really something I'd thought about. Why does it happen? What determines the shape of this glare and how big it is.

    Glare comes up in several topics here in Metabunk, but there's two in particular. Firstly there's the glare around the Sun, which has led some Flat Earth people to claim that the sun gets visibly bigger during the day. However it actually stays the same size and is just brighter.

    Secondly, there are UFOs on thermal cameras. Jet engines are very hot, and the glare around those engines can be bigger than the plane itself, resulting in an off shape. The classic example is the Chilean Navy UFO
    [​IMG]
    Here the actual "light" (the heat source) is quite small, but the glare is very large. Multiple glares combine together and overlay, obscuring the actual shape of the planes.

    That example was proven by radar tracks to be that particular plane. A hypothesized example is the "Gimbal" UFO video:
    [​IMG]

    This was discussed at some length in that thread and continues to be discussed on Facebook. The notable things about the Gimbal video are:
    1. The rotation, which seems very unlike a normal aircraft.
    2. The shape of the glare - a flatter oval, with protuberances
    3. The "aura" around the object (which is actually a darkening, as the video is black=hot)

    #1 and #3 are discussed at length in the other thread. Both things can be seen in this animation by @igoddard :

    [​IMG]

    The aura is simply something you get around any bright glare in an IR image. It does not indicate anthing of interest.

    The rotation of the glare is due to the rotation of the camera relative to the horizon. The image is then derotated to keep the horizon level with the plane. Since the glare is an artifact of the optical system it rotates relative to the horizon, even when there's nothing else rotating. This derotation is described in the patent for the ATFLIR system used for the GIMBAL video:
    [​IMG]
    This derotation, and the potential for rotating glare/flare is discussed here. It's not really disputed, but can be difficult to explain. However it's clear that rotating glares can happen.

    Glare or Flare
    I'd previously referred to this as a "flare" or "glare" interchangably. However "flare" is more commonly associated with "lens flare", which is a reflection inside the lens resulting in multiple images of the light - often as a line of different circles, or sometimes as a single reflection. To avoid confusion I'm going to try to stick to "glare" - meaning an enlarged and distorted shape of the light viewed directly.
    Metabunk 2019-03-30 08-43-13.


    So with the rotation and aura explained, the question remaining with the Gimbal video is: "why is it shaped like that?" and "what is the underlying object."

    The first question raised would be "is it even a glare?" Perhaps instead this is the actual shape of the object?
    Metabunk 2019-03-30 08-53-21.

    I think not, largely because of the way the shape changes. Is the above the true shape, or is this early shape?
    Metabunk 2019-03-30 09-02-16.
    Of the more dramatically different "white hot" image at the start?
    Metabunk 2019-03-30 09-04-03.

    But perhaps the biggest problem with this hypothesis is: how do we get a glare with this shape? In the Chilean Navy case, the shape of the glare came from the configuration of the engines. But here the shape rotates, which suggests it is an artifact of the optical system. Can we get saucer-shaped glares that rotate? Yes, we can:

    20171219-100429-rpsp1.

    This again was discussed in the other thread. A flashlight with the reflector removed was used a bright light source. Images were taken with an infrared camera at different angles, and the result was a saucer shaped flare with a distinct long axis that roated based on the camera.

    Metabunk 2019-03-30 09-15-20.

    This is nothing new, but with the renewed discussion on Facebook I set out to investigate more what actually makes these shapes, and if the Gimbal shape was plausible.

    For one experiment I set up a bright light and videoed it with a camera that I rotated by hand, then derotated in software (the equivalent of the ATFLIR mirror derotator)
    Flashing and Contour HD roating flare setup.

    The result again was a saucer-shaped glare that rotated.

    via GIPHY


    Source: https://giphy.com/gifs/1msBbaccpCov94oWlh/html5


    Now, this is still a ways from replicating the Gimbal glare shape, but it's a start. Given the complexity of the gimbal optical path, it seems plausible a more complex shape would emerge.

    Something that struck me was that I consistently got this saucer shape with different cameras. My initial idea was that the glare was due to streaks on the glass at the front of the ATFLIR. This is probably what leads to rotating glare like this:
    http://www.military.com/video/opera...trikes/f-18-takes-out-insurgents/658386321001
    [​IMG]

    Streaky glass tends to give these long flare though, which is not what we see. While I did replicate a rotating saucer-shaped glare by rotating streaked glass, it still had some long streaks.

    Metabunk 2019-03-30 09-48-05.
    Note though that it also had some short streaks, which kind of matches the short axis of the Gimbal:
    [​IMG]
    So it's possible that the streaks we just too thin to be recorded in the video.

    This whole issues kind of got lost in the other thread, so I'd like to revive (and hopefully resolve) it here. What could cause this rotating glare shape? Can it be replicated?
     
    Last edited: Apr 5, 2019
  2. Mick West

    Mick West Administrator Staff Member

    via GIPHY


    Source: https://giphy.com/gifs/2cdYISoU7GZA3uxhki/html5


    Here I've adjusted the video of my "streaky slide" to be more like the Gimbal video in terms of contrast and resolution. The long streaks are less apparent. It does change shape a lot, but that's probably because the slide is moving around, whereas in the Gimble situation it would be more fixed.

    However, going back to this one:

    via GIPHY


    Source: https://giphy.com/gifs/1msBbaccpCov94oWlh/html5


    There was no extra streaky glass. So where does the shape come from? Well, perhaps from the glass on front of the camera? It's a bit dirty, not really streaky though.
    Metabunk 2019-03-30 10-22-03.

    I gave it a thorough cleaning, and this did seem to reduce the oblateness of the glare.
    Metabunk 2019-03-30 10-29-30.

    Hard to get REALLY clean though. This is after I cleaned it.
    Metabunk 2019-03-30 10-42-29.
     
    Last edited: Mar 30, 2019
  3. Gerard

    Gerard New Member

    I have some questions about this experiment.

    Was this an IR camera ? Otherwise I'm wondering why the light appears black.

    What causes the small secondary black spot that is moving around ? - lens reflection ?

    What does the "derotating" software actually do ? Does it track image features to infer a rotation matrix which it then inverts ?

    It might be helpful to see the un-derotated video to get a better sense of what the software is doing.
     
  4. Mick West

    Mick West Administrator Staff Member

    It's IR, (actually this one is not, the other one with the slide is IR) but the image is inverted, to match the Gimbal video (which is mostly inverted, but starts out uninverted, with white=hot)

    Lens reflection, a normal "lens flare" (not glare)

    Essentially yes. See the two yellow balls? I put them there to give the software something easy to track.

    via GIPHY


    Source: https://giphy.com/gifs/fs0roV8TIy5kOGu5Ha/html5
     
  5. Mick West

    Mick West Administrator Staff Member

    Sorry, just to clarify this one (in color) is obviously not IR. I just converted it to B&W and inverted. The video with the slide is using an IR camera but is also inverted.
     
  6. Gerard

    Gerard New Member

    I think the effect is even clearer without the derotation. You can clearly see the glare move relative to the background, which I guess proves that it's coming from the lens or something attached to the camera rather than from the light.
     
  7. Mick West

    Mick West Administrator Staff Member

    I think to get into the details here we need to understand what causes glare in the first place. Why is there a big light in the image when the actual light is small? The answer seems to be mostly diffraction.

    When the light from a bright object makes an image on the sensor (or film) then theoretically all the light should fall on that image. However whenever light hits something, like the edge of the aperture, or dirt or streaks on the glass, or imperfections in the lens, then it gets bent by diffraction, which spreads out the light in directions that depend on the shape of the obstruction.

    In astrophotography, large telescopes have a setup like this:
    [​IMG]

    Which leads to what are called "diffraction spikes", like in this image from the Hubble Space telescope.
    [​IMG]


    You also get less dramatic spikes from apertures that are not circular. The degree of the spike is related to the size of the aperture. Here's one photo at F/4.5
    Metabunk 2019-03-30 11-15-27.
    F/10
    Metabunk 2019-03-30 11-16-34.

    And then F/20
    Metabunk 2019-03-30 11-16-00.

    Now in all three cases there still significant glare, but smaller spikes with the wider aperture. The wide open glare seems more circular, and the small aperture glare is more hexagonal.

    So it would seem that the inner glare here is the even diffusion of light all around the edge of the aperture (or just the circular sides of the lens interior if the aperture is fully open).

    This glare is lens relative, and so it does the same kind of rotation we are talking about here.

    So, to extend this to the Gimbal video, could it be something in the camera itself making this shape? Some diffraction spikes from something in the light path, like the de-rotating mirrors? Or maybe a combination with dirt/water/smudges on the lens?
     
  8. Gerard

    Gerard New Member

    In the case of the Gimbal video, isn't the purpose of the camera rotation mechanism only to counteract changes in the aircraft's attitude ?

    If that's the case it shouldn't be moving unless the aircraft's attitude is changing. I think there's HUD symbology in the FLIR videos which indicate the aircraft attitude. Can we tell if the apparent object rotations correlate with changes in that symbology ?
     
  9. Mendel

    Mendel Member

  10. Mick West

    Mick West Administrator Staff Member

    No, it has to rotate the view because there's only two axes of rotation of the ATFLIR, one of which is along the long axis, so you can't track something without rotating the view. See this example:

    Source: https://www.youtube.com/watch?v=fxlKkn1b4IY


    So the derotation is to correct this "pendulum effect". It's also a bit more complicated as there are coarse and fine gimbals, so the derotation need not be continuous.
     
  11. Mick West

    Mick West Administrator Staff Member

    Much more of an issue for the Sun, I'd think. I'd be very surprised if atmospheric glare was a factor here. Particularly as it would not rotate.
     
  12. Gerard

    Gerard New Member

    I think it would be Mie scattering from particles or imperfections on the lens.
     
  13. Mick West

    Mick West Administrator Staff Member

    Glare Generation Based on Wave Optics - Kakimoto, et al,
    Metabunk 2019-03-30 12-44-55.
    http://nishitalab.org/user/nis/cdrom/pg/glare_m.pdf
    While this article is focussed on the eye, it confirms the glare is largely a diffraction phenomenon from obstacles and the edges of the aperture (the pupil in this case).
     
  14. Mick West

    Mick West Administrator Staff Member

    Some relevant things from the longer thread:
    Following that link we see it talks about "veiling glare" which is a general term for glare that obscure part of the image, and seems frequently to refer to glare that gives the entire image a hazy look with reduced contrast.
    Metabunk 2019-03-30 12-57-47.

    In the "Glare Spread Function" chart. The narrow part of the central spike seems to be the actual sun, the widening at the "base" is perhaps hard glare then the light below this is much less.
     
  15. Mick West

    Mick West Administrator Staff Member

    To investigate the shape of glare, I set up an open "camera" consisting of a lens, and a backplane representing where the sensor would be.
    Metabunk 2019-03-30 14-23-14.
    This was put in the sun pointing at the window. You can see the sun on the paper, and a bit of the window. The lines are used for focus.
    Metabunk 2019-03-30 14-24-33.

    There's the glare! Or is it? I found it hard to tell what was light spread on the projected image, and what was glare in the camera taking a photo of the "photo"

    I set up a bigger camera, at an angle Metabunk 2019-03-30 14-27-18.
    I then used a variable ND filter to adjust the brightness, getting

    via GIPHY


    Source: https://giphy.com/gifs/KY5c7OukX8JI6VPcOh/html5


    I then moved the camera so it was at a shallow angle, and the image flattened, showing the flare was from the first lens, and what we see in the paper is what actually arrives on the sensor (for this lens, at that aperture)

    via GIPHY


    Source: https://giphy.com/gifs/3b8Mo076AN67THvgsh/html5
     
  16. Mick West

    Mick West Administrator Staff Member