Replicating the nvidia light field displays - lenses?

[Apropos]

Many of you have probably seen the nvidia light field display at siggraph. In case you missed it:

http://www.youtube.com/watch?v=uwCwtBxZM7g

http://www.youtube.com/watch?v=kcmD2s6E-LE


Anyway I think this was my reaction This is what I think will be a major future technology.

That's why I'm trying to replicate it. For static images it's quite easy to generate the tiled image required for lightfield generation from a 3d scene. The problem I ran into is sourcing the lenses.
According to the research PDF: https://research.nvidia.com/publication ... d-displays the lenses are

Microlens arrays were affixed to the displays, weighing 0:7 grams and having a
1:0 mm lens pitch and 3:3 mm focal length.

I haven't been able to find anything like this and no optics shop around here wants to manufacture them for me. Can anybody help me?

[android78]

One way you may be able to do this on the cheap is using a series of pinhole lenses. I haven't tried it, but my thought was to print on transparency with mostly black, but white dots for the pinholes.
I wish I had bookmarked the page now, but I found the page from a guy who had made his own using bearing balls. His way was quite complicated though with milling holes in which to place balls, etc, and I think you could get reasonable results just by putting the balls in a container of correct dimensions to hold them in pattern and then just pour de-gassed silicone to make a mould. Once you have that, you would just cut the silicone at the balls mid-point.

EDIT: Found it:
http://lightfield-forum.com/2012/05/how ... ld-camera/

[Apropos]

That is brilliant.

The Ball Bearing technique will work for square lenses but not for 14x8 lenses. It is something worth trying though.

The pinhole technique is intriguing but the problem I see right off the bat is that the image will be upside down. This wouldn't be such a big problem but it does have to be accounted for.
Secondly printed transparencies are not that good at blocking light. I know from my experiences with PCB masks made from transparencies. You'd need 3 layered to have a good enough blockage.
Both those problems accounted for I would try it right now but by my calculations the hole size should be 0.077mm. No that's not a typo. That means at least a 2400DPI printer with a 20 micron dot size should be used to trace the holes. Not out of reach for us but that's double the res of my printer. It might be easier to print on a photo transparency http://en.wikipedia.org/wiki/Color_transparency_film

[FMPrime]

Check out the work done at MIT media lab.
http://displayblocks.org/diycompressive ... rdisplays/

They have combined the lenslet array, parallax barrier, and active LCD layers technologies to greatly increase the resolution of glasses-free 3D.

I think these displays in a rift would be the ultimate.

[profvr]

The lens used is #630 from Fresnel Technologies. $40/sheet, but $100 minimum order (at least if you are in the EU).

http://www.fresneltech.com/visible.html


ProfVR

[geekmaster]

... I found the page from a guy who had made his own using bearing balls. ...
EDIT: Found it:
http://lightfield-forum.com/2012/05/how ... ld-camera/

More info (with pictures) here:
https://developer.oculusvr.com/forums/v ... 049#p35589