2
\$\begingroup\$

I am trying to construct a display that is shaped like an eighth piece of a sphere. The resulting device shall be self-sufficient and possibly portable, meaning projection onto matte octant won't do it. The screen should be capable of displaying photorealistic images at a high frame rate, so POV display won't suffice as well. Primarily, I see two options here:

  • to curve light output from a regular “planar” display with a lens or anyhow else,
  • or to give needed curvature to the display matrix itself.

Having thought over the first approach, I foresee two main concerns to deal with. The first one is that homogenous pixel density will become heterogenous when light is bent in non-isomorphic manner (see figure).

Planar display becomes spherical, scale exaggerated

That is, to make a plane into a sphere, we need to stretch periphery more than centre, therefore pixel density will be affected likewise so we get fewer pixels to corners. This could be partially mitigated through software, but the problem still remains. The second problem is the lens itself. I am not confident enough on the right branch of optics or optical engineering to be able to determine which type of optical device and of what metrics should be used here.

The second mentioned approach with already curved matrix seems less complicating, although I do not know how to produce such displays, or how to order them; and I am not sure about image quality of such displays, should they exist.

Thank you for your interest. Any suggestion would be much appreciated.

\$\endgroup\$
  • \$\begingroup\$ Many questions: How big is this? How close to uniform do you want the pixel density to be? Do you intend to make 8 of these and assemble them into a full sphere? (This would put certain constraints on the possible implementations.) Must it be self-contained (lit from behind), or could you simply project an image onto a white octant from the outside? \$\endgroup\$ – Dave Tweed Nov 1 '14 at 14:52
  • \$\begingroup\$ Thank you for useful questions, Dave. Size is meant to be of order of magnitude of a laptop screen; twenty centimetres of radius or so would suffice. I want it to be so that the average human eye could not distinguish separate pixels from viewing distance. Putting eight of them together would be ideal in the future, yet that is a different story — I need to prove the concept first. Yes, it should be self-contained. I've updated the post accordingly. \$\endgroup\$ – DJ HiFi Nov 1 '14 at 18:09
1
\$\begingroup\$

I think that given the constraints on size and resolution, interior projection using some form of Pico projector will probably be your best approach. The DLP-based units from TI are particularly nice.

You'll probably need to find a way to fold the optical path using one or two mirrors in order to make it self-contained, while still covering the entire octant.

\$\endgroup\$
3
\$\begingroup\$

Have you heard of POV (persistance of vision) displays? This video shows such a display in action, it is a circle of LEDs that is spun around with the LEDs being controlled depending on rotational position. You could do the same, but switch off the display everywhere but that one eigth of the sphere you want to use.

Another option would be to project the image onto the inside (or outside) of a white, thin piece of spherical plastic or similar, using a simple projector. In this case, you would of course have to run a bit mathematics on the image before displaying it in order to not stretch it.

\$\endgroup\$
  • \$\begingroup\$ Thank you for the suggestions. Unfortunately, I have never encountered a POV that is able to deliver pictures realistic enough for my task. Projection from inside, on the other hand, seems quite promising. Could you please point me at a prior implementation of such a device? \$\endgroup\$ – DJ HiFi Nov 1 '14 at 18:14
  • 1
    \$\begingroup\$ A quick search turned up this: makezine.com/2011/09/19/… \$\endgroup\$ – QuantumFlux Nov 1 '14 at 22:38
  • \$\begingroup\$ Also, you might want to update your question to say that you want a somewhat high resolution, that way other people might be able to answer more specifically. \$\endgroup\$ – QuantumFlux Nov 1 '14 at 23:02
  • \$\begingroup\$ It's embarrassing how I missed this. Looks like Nirav Patel's work is indeed what I want. \$\endgroup\$ – DJ HiFi Nov 2 '14 at 6:54

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.