Forum Discussion

🚨 This forum is archived and read-only. To submit a forum post, please visit our new Developer Forum. 🚨
romacron's avatar
romacron
Honored Guest
12 years ago

How far is the focal plane away from the eyes?

Has anyone an idea how far the focal plane is away from the eyes?

7 Replies

  • DiCon's avatar
    DiCon
    Honored Guest
    I am not sure, if I get your question right, but let's try...

    In the ideal/simple case, the focal plane is on the screen. The light emitted from the screen is parallel after passing the lens, so that your eyes have to focus on "infinity" - then the lenses of your eyes focus the light on your retina.

    So... If you are asking for the focal length as a property of the lenses, the answer is that it's the distance of the screen to the lens. You can easily find this distance by holding your lens a few cm above the floor and change the distance until you can see a perfect projection of your ceiling lights: If there are multiple light sources on your ceiling (like tubes or multiple bulbs), you can see each one distinctly on the floor below your lens - the distance from the floor to the lens is (almost) the focal length as the light from a far light source is (almost) parallel.

    If you are asking for the planes in which the actual light is focussed while using the Rift, the answer is simple: On the screen and on your retina (which is actually done by the lenses of your eyes) - otherwise it would not work. You can also notice, that there is no additional focal plane in-between as the image on the screen is not flipped when you remove the lenses and inspect it directly.

    EDIT: Removed stuff about relief setting - That was wrong - just checked it and the distance between lenses and screen does not change at all - makes these explanations much simpler :)
  • "romacron" wrote:
    Has anyone an idea how far the focal plane is away from the eyes?


    If you think in terms of a traditional stereoscopic convergence plane? For a VR HMD that would be placed at infinity. HMD stereoscopy is fundamentally different to fixed screen stereoscopy. In a fixed screen setting, the screen is effectively a window the user is looking through, onto which the eyes will involuntarily converge onto the same viewport and you have to align your projection with that. In an HMD however there is no "window"; each eye has its own viewport.

    Because of that there's no parallax shift in a HMD projection model whatsoever. The projection axes are parallel.
  • A bit more clarification on what you're looking for might be in order, perhaps. But I think Datenwolf nailed it, if you're creating stereo images/video/etc to be view in the Rift. Long story short, set it up with parallel cameras.

    That said, creating 3D content for the Rift is a bit different than traditional 3D content. I've made a number of 3D projects, always using a wide variance of convergence (i.e. "screen depth", like Datenwolf talked about), so the idea of making something without convergence seems counterintuitive to me, but I also know that's how it has to be done.

    Which just now makes me realize that my Rift will be here in a few days... I should get cracking on making some basic 3D stills to wow myself with.. lol.
  • It's not really counterintuitive once you've peeked behind the projective math of it all. In the Rift you effectively got two screens (yes I know it's one large widescreen, but that doesn't matter right now) where each eye, when looking into infinity, looks dead-center onto the screen. This means you've got a vanishing parallax shift. In traditional stereoscopy however you've got a shared viewport and you have to shift the projection plane so that on the projection screen the projection of a single point at infinity toward the eyes is at about eye separation in the screen plane.

    A lot of people get the whole stereoscopy theory wrong, because they start of the assumption of our eyes "sverving in" to a convergence point and you should resemble that. But this is wrong. What you really have to do is a back projection of the offset of between the eyes' pupils to the screen center. And for a HMD that value is 0.

    BTW, if you want to look at my stereoscopy work have a look at this: http://orange.blender.org/blog/elephants-dream-in-stereoscopic-3d/
  • j1vvy's avatar
    j1vvy
    Honored Guest
    Thanks for the info. I have been making some S3D panoramic videos for almost a year now but only recently got a Rift. I would test the alignment using VRPlayer and anaglyph glasses to move the projection plane. I did this by rotating the right view 4.5° so instead of the aligning at the horizon it would be aligned about 2 to 3 feet in front of the camera.

    I notice that know that I have a Rift when I view in VRPlayer I have to add a horizontal shift to get the views to converge.

    So does that mean that when I view any stereoscopy movie made for watching on a TV I will have to shift as well?

    Because of a bug in VRPlayer it is not possible to shift the view when viewing as anaglyph and by aligning to the horizon everything will look like it is coming out of the screen.

    I don't really want to have to create multiple versions.

    I down loaded Elephants Dream but it ended up being a 2D version. 3D version dl now.
  • INOV8's avatar
    INOV8
    Honored Guest
    HI J1vvy

    What type of camera unit are you using for your panorama 3d videos.

    Thanks

    D
  • j1vvy's avatar
    j1vvy
    Honored Guest
    I created a rig using 7 overlapping pairs of inexpensive Mobius cameras to capture the horizon then two more to fill in the zenith in 2D. The examples will be updated to fix the L&R alignment in days to come.
    http://photocreations.ca/3D/mobius_camera_rig.html

    I will be upgrading to lenses with wider angle of view so I can have fewer seams.