Welcome to PhantomPilots.com

Sign up for a weekly email of the latest drone news & information

Mavic Focus

Discussion in 'Mavic Discussion' started by datincan, Oct 2, 2016.

  1. datincan

    Joined:
    Sep 28, 2015
    Messages:
    68
    Likes Received:
    5
    Is the focus remaining fixed once it is locked onto a subject?

    i.e. does it track the subject to maintain focus - or if the mavic moves away from a fixed subject will it loose focus?
    for example, one of the demo shots on a review showed a pull away shot from the top of a small light house. - If the camera was first focused while close to the light house and as the mavic pulls up from it, will the light house and all subjects in similar focal plane become out of focus? ...or will the camera track the lighthouse and keep it in focus all the time?
     
    #1 datincan, Oct 2, 2016
    Last edited: Oct 2, 2016
    YassPro and Big Dog like this.
  2. Big Dog

    Joined:
    Feb 17, 2016
    Messages:
    170
    Likes Received:
    59
    Good question.


    Sent from my iPhone using PhantomPilots
     
  3. hudster

    Joined:
    May 15, 2016
    Messages:
    100
    Likes Received:
    15
    Location:
    West Pittston, Pa
  4. datincan

    Joined:
    Sep 28, 2015
    Messages:
    68
    Likes Received:
    5
    I guess we will in time, I was hoping that someone who already has one could confirm this now thought..
     
  5. TonG

    Joined:
    Aug 14, 2014
    Messages:
    294
    Likes Received:
    74
    Location:
    The Netherlands
    The hyperfocal distance from this lens (28mm equiv. and F2.2) is around 12 meters.
    That means if you focus on infinity that the minimal distance a subject will be sharp is 12 meters .
    The reverse is the same; focus on 12 meter and you have sharpness to infinity.
     
    datincan likes this.
  6. DarkEnergy

    Joined:
    Sep 24, 2016
    Messages:
    23
    Likes Received:
    6
    Location:
    Cambridge, UK
    I don't think the reverse is correct. If you focus at 12m the focus region (depth of focus) will start at a distance below 12m and will not reach infinity.
     
    Jeiel Shamblee likes this.
  7. TonG

    Joined:
    Aug 14, 2014
    Messages:
    294
    Likes Received:
    74
    Location:
    The Netherlands
    Thats an irresistible logic.
    I might be wrong! I have to figure it out. Thanks


    A few minutes later.....
    I was partly wrong. The hyperfocal distance is 12m (11.7m) but the near limit of sharpness is at 5.92m.
    So focus on 12m and the far limit is at infinity. Focus on infinity and the near limit is 12 meter.

    Schermafbeelding 2016-10-03 om 21.52.54.png
     
    #7 TonG, Oct 3, 2016
    Last edited: Oct 3, 2016
  8. DefaultIT

    Joined:
    Jan 23, 2016
    Messages:
    188
    Likes Received:
    74
    I think actually the reverse is the one that's true, if calculated correctly; Hyperfocal = minimum focal distance at which infinity is still acceptably sharp. (EDIT oops meters...yeah about 39ft) Setting the camera to hyperfocal gets you a depth of field from about half that distance to infinity.

    This is all made a bit fuzzy (badump) by the definition of "acceptably sharp" which is resolution-dependent. Also it's a bit of an academic exercise since there doesn't appear to be any focus distance indication from the camera itself; you'd have to pick something about the right distance away and hope the electronic contrast detection does a reasonably accurate job of adjusting the lens to it.
     
  9. TonG

    Joined:
    Aug 14, 2014
    Messages:
    294
    Likes Received:
    74
    Location:
    The Netherlands
    That means if you focus on infinity you get sharpness way beyond infinity. Wow!:)
     
  10. DefaultIT

    Joined:
    Jan 23, 2016
    Messages:
    188
    Likes Received:
    74
    Indeed....Buzz would be proud
    [​IMG]
     
    TonG likes this.
  11. DarkEnergy

    Joined:
    Sep 24, 2016
    Messages:
    23
    Likes Received:
    6
    Location:
    Cambridge, UK
    I can see now that there is a problem with your estimate of the FOV
    Firstly the diameter of circle of confusion seems too large. In a Nikon D4 corresponds to 4 pixels in diameter.
    Secondly, your computation is for a full frame camera (36x24mm) while the Mavic has a 1/2.3" detector that is about 6.2x4.6mm and roughly the same number of pixels (well 16M of the Nikon vs 12M of the Mavic).
    A 0.03mm diameter circle of confusion in the Mavic 1/2.3" detector, corresponds to about 20 pixels if I'm not wrong.This is far too large.
    Based on the sampling theorem (Nyquist-Shannon) and to preserve resolution, the diameter of the circle of confusion in a digital detector should be about 2.5 pixels or in the case of the 12Mpx 1/2.3" detector, about 0.004mm. If you repeat the calculation using this value for the circle of confusion you will see why autofocus is necessary. Clearly all this is assuming that Mavic's optics can deliver optimum images.

    Screen Shot 2016-10-04 at 09.27.20.png
     
  12. TonG

    Joined:
    Aug 14, 2014
    Messages:
    294
    Likes Received:
    74
    Location:
    The Netherlands
    The subject is much more complicated and interesting topic then I thought.
    I looks that you are not right. If you take the smaller CoC you should take the original focal length.

    However... I did a quick and dirty test with my P3P.
    With its fixed focus (20mm eq. F2.8) I got a sharp image from 1m70 and further to infinity.
    If I use your calculations this is impossible.
    If I use a focal length in the calculator of 5mm the outcome looks very alike what I see in my testpicture.

    DOF-calc.png
     
  13. DarkEnergy

    Joined:
    Sep 24, 2016
    Messages:
    23
    Likes Received:
    6
    Location:
    Cambridge, UK
    Yes you are correct, my mistake.
     
  14. DarkEnergy

    Joined:
    Sep 24, 2016
    Messages:
    23
    Likes Received:
    6
    Location:
    Cambridge, UK
    Just to settle the point, I've made an estimate of the focal length of Mavic's camera based on the size of the 1/2.3" detector and the claimed FOV of 78.8ยบ.
    If the FOV corresponds to the diagonal of the detector, as is commonly quoted in photography, the true focal should be about 9.4mm. For a focal length of 9.4mm, f-ratio of 2.2 and CoC=0.004mm, the hyperfocal distance is about 10m, similar to your early result.


    Screen Shot 2016-10-04 at 11.50.59.png Screen Shot 2016-10-04 at 11.46.29.png
     
  15. David_Cambridge

    Joined:
    Jul 29, 2016
    Messages:
    70
    Likes Received:
    14
    All,

    Circle of confusion is also impacted by the viewing distance of the resulting image. The eye itself has a resolution limit. If the image is moved closer to the eye, increased image resolution is needed to be percieved as sharp. Conversely, moving the image further from the eye, past the point of the specific eye's resolution, one cannot seethe full sharpness in the image.

    Individual eyes have significant differences in resolution, and eyes change their resolution as they age. The gold standard for measuring image resolution is measuring the modulation transfer function between alternating black and white lines. When the contrast between the light and dark line falls to 50% of it's initial value, it is often considered out of focus. There is literally only a single infinitely thin plane of perfect focus.

    The psychovisual use of circle of confusion is an abstraction from this to what a normal eye sees.


    Sent from my iPad using PhantomPilots
     
  16. WetDog

    Joined:
    Oct 18, 2015
    Messages:
    1,491
    Likes Received:
    640
    Indeed. Traditional COCs were calculated for a print (can't remember the size) at a 'standard' viewing distance. This has not so much to do with how we typically look at an image these days. The calculated COC of 0.004 mm is probably a good one to use for general purposes.
     
  17. Peter Galbavy

    Joined:
    Aug 25, 2016
    Messages:
    66
    Likes Received:
    26
    I'll also throw in not knowing the Bayer layout of the sensor (the traditional R-G-G-B is not always the case for video oriented sensors) and hence the CoC can only be an estimate. It was only ever an estimate for film anyway.
     
  18. DarkEnergy

    Joined:
    Sep 24, 2016
    Messages:
    23
    Likes Received:
    6
    Location:
    Cambridge, UK
    You are confusing eye resolution with the photographic camera or video camera resolution. Regarding the modulation transfer function you mention, that is I believe another way of expressing the sampling theorem AKA the Nyquist-Shanon theorem mentioned above in reply 11.
     
  19. David_Cambridge

    Joined:
    Jul 29, 2016
    Messages:
    70
    Likes Received:
    14
    No.

    I am attempting to separate an d clarify the difference between them. If you are attempting to compare the Mavic's resolution and a Phantom, the way to do that is with the modulation function method. That will return a measurement that resolves the comparison exactly.

    When you start discussing visually resolution, which is our human perception of vision, then the circle of confusion comes into focus
     
  20. DarkEnergy

    Joined:
    Sep 24, 2016
    Messages:
    23
    Likes Received:
    6
    Location:
    Cambridge, UK
    The CoC of a given photographic camera or a video camera has nothing to do with human perception, it is a property of the camera. You can analyze the image with your naked eye or a magnifying glass or even a microscope and the result will always be the same.