Searching \ for '[OT] 3D display uwing spinning mirror and high fra' in subject line. ()
Make payments with PayPal - it's fast, free and secure! Help us get a faster server
FAQ page: www.piclist.com/techref/io/serial/spis.htm?key=spi
Search entire site for: '3D display uwing spinning mirror and high fra'.

Exact match. Not showing close matches.
PICList Thread
'[OT] 3D display uwing spinning mirror and high fra'
2010\01\23@035709 by Peter

picon face
www.jimonlight.com/2009/09/04/a-rotating-autostereoscopic-display/

I believe that a simlar system could be used to project a small 3D image from a
LED dot matrix display driven by a microcontroller. The required processing
power would be in range for a PIC to keep up.

 Peter

2010\01\27@124700 by M. Adam Davis

face picon face
Find LEDs that have a 1.2 degree horizontal and 180 degree vertical
viewing angle, and then just spin an array of them around.  In fact,
skip the LEDs and just spin an LCD around with a mask to limit the
viewing angle horizontally.

Unfortunately you don't get vertical seperation for any of these
devices, and cheap displays are bound to be noisy.  If you want 30
frame per second video you need to spin the display at 1,800 RPM.

-Adam

On Sat, Jan 23, 2010 at 3:56 AM, Peter <spam_OUTplpeter2006TakeThisOuTspamyahoo.com> wrote:
> http://www.jimonlight.com/2009/09/04/a-rotating-autostereoscopic-display/
>
> I believe that a simlar system could be used to project a small 3D image from a
> LED dot matrix display driven by a microcontroller. The required processing
> power would be in range for a PIC to keep up.
>
>  Peter
>
>

2010\01\28@063828 by Isaac Marino Bavaresco

flavicon
face
Em 27/1/2010 15:46, M. Adam Davis escreveu:
> Find LEDs that have a 1.2 degree horizontal and 180 degree vertical
> viewing angle, and then just spin an array of them around.  In fact,
> skip the LEDs and just spin an LCD around with a mask to limit the
> viewing angle horizontally.
>
> Unfortunately you don't get vertical seperation for any of these
> devices, and cheap displays are bound to be noisy.  If you want 30
> frame per second video you need to spin the display at 1,800 RPM.
>
> -Adam
>  

And the display needs to show thousands of frames per second, because
you need to change the image lots of times each revolution.

Regards,

Isaac

__________________________________________________
Faça ligações para outros computadores com o novo Yahoo! Messenger
http://br.beta.messenger.yahoo.com/

2010\01\28@070700 by Peter

picon face
M. Adam Davis <stienman <at> gmail.com> writes:
> Find LEDs that have a 1.2 degree horizontal and 180 degree vertical
> viewing angle, and then just spin an array of them around.  In fact,
> skip the LEDs and just spin an LCD around with a mask to limit the
> viewing angle horizontally.

I am not so sure that the narrow angle is needed. Their algorythm seems to
deliberately apply cos^2 horizontal shading to each frame. This makes perfect
sense if you think about it. The slow motion images when spinning up in the
video seem to show such shading.

For a spinning led matrix the viewing angle would actually become wider, not
narrower, if the shading hypothesis is true.  What I do not yet understand is
how the viewer rising and lowering his/her head changes the perspective in the
vertical plane. It seems that the answer to that question is 'no, the viewer
cannot change the perspctive by rising or lowering his viewpoint' for now but I
may be wrong.

I make an educated guess saying that a simple limited 3d one-line led based
propeller display could easily be modified to use this 3d principle, by simply
placing the line of leds at an angle (think line diagonal in a cylinder). This
should be able to display any biconical shape.

Similarily spinning a standard upright 5x7 matrix display could display any
image that fits in a cylinder with r=3 and h=6. The only reason the mirror is
not upright in the original video, is that the image source is above the case. A
different, older type of the same display as shown in the video uses a spiral
mirror (one turn, one blade propeller with pitch equal to the cylinder height).
There are many variations possible on this theme.

> Unfortunately you don't get vertical seperation for any of these
> devices, and cheap displays are bound to be noisy.  If you want 30
> frame per second video you need to spin the display at 1,800 RPM.

The bright side of it is that one can make a closed sphere or hemisphere topped
bell and spin the whole of it, with the mirror or led plane an oval internal
plane section of it. That would eliminate the 'fan' effect seen in propeller
type displays (but some pumping effect would remain, based on Magnus and Tesla
effects from the outer transparent spinning round body). 1800 rpm is the default
speed of most 2 phase ac motors (shaded pole, capacitor run etc). Obtaining 1800
rpm is not hard for reasonably small spun objects.

 Peter

2010\01\28@072910 by Dave Tweed

face
flavicon
face
Peter wrote:
> M. Adam Davis <stienman <at> gmail.com> writes:
> > Find LEDs that have a 1.2 degree horizontal and 180 degree vertical
> > viewing angle, and then just spin an array of them around.  In fact,
> > skip the LEDs and just spin an LCD around with a mask to limit the
> > viewing angle horizontally.
>
> I am not so sure that the narrow angle is needed. Their algorythm seems
> to deliberately apply cos^2 horizontal shading to each frame. This makes
> perfect sense if you think about it. The slow motion images when spinning
> up in the video seem to show such shading.
>
> For a spinning led matrix the viewing angle would actually become wider,
> not narrower, if the shading hypothesis is true.  What I do not yet
> understand is how the viewer rising and lowering his/her head changes the
> perspective in the vertical plane. It seems that the answer to that question
> is 'no, the viewer cannot change the perspctive by rising or lowering his
> viewpoint' for now but I may be wrong.

You missed a couple of key points in that demo. First of all is the
anisotropic mirror, which is quire specular in the horizontal plane,
sending the CRT image in a particular direction, while scattering light
broadly in the vertical plane. The display will be quite blurry if you
can't constrain the distribution of light in the horizontal plane. In
particular, you won't be able to do occlusion.

Head tracking is used to simulate parallax in the vertical dimension.
They didn't show it in the video, but this aspect of the display only
works for one viewer at a time.

-- Dave Tweed

2010\01\28@073012 by Alan B. Pearce

face picon face
>I make an educated guess saying that a simple limited 3d
>one-line led based propeller display could easily be modified
>to use this 3d principle, by simply placing the line of leds
>at an angle (think line diagonal in a cylinder). This
>should be able to display any biconical shape.

I would have thought that the way to do it with a propeller clock type
emitter would be to have a head sensing trick like the guy that uses a Wii
handheld sensor strapped to his head for sensing to rotate the image on a TV
screen.

2010\01\28@083905 by Peter

picon face
Dave Tweed <pic <at> dtweed.com> writes:
> You missed a couple of key points in that demo. First of all is the
> anisotropic mirror, which is quire specular in the horizontal plane,
> sending the CRT image in a particular direction, while scattering light
> broadly in the vertical plane. The display will be quite blurry if you
> can't constrain the distribution of light in the horizontal plane. In
> particular, you won't be able to do occlusion.

Interesting, are you saying that the mirror is special? A normal mirror would be
*very* specular in all directions and essentially show an image to a viewer for
only one frame. Are you saying that the mirror is anisotropic?

> Head tracking is used to simulate parallax in the vertical dimension.
> They didn't show it in the video, but this aspect of the display only
> works for one viewer at a time.

I see. That makes sense.

 Peter




2010\01\28@092555 by Peter

picon face
What screen display technology is suitable for a frame rate of 4000+fps ?! An
electrostatically scanned oscilloscope tube could do it technically. The video
rate would be a scary minimum of 300MHz for 200+ frames/rotation and the
computer video adapter and driver supplying this data stream would be of
'interesting' design. I assume that one can render into minibuffers in video ram
and switch the buffers rapidly (one per frame). 800*600*3bytes/pixel would fit
in 1.44MB and fitting 300 of those mini buffers in a contemporary 512MB+ video
card is possible.

Peter


More... (looser matching)
- Last day of these posts
- In 2010 , 2011 only
- Today
- New search...