Silicon-based LED arrays the future of VR/AR displays?

[MSat]

I had mentioned before that since LEDs made on silicon substrate are about to reach the market in a year or so, all the pieces needed to make an LED microdisplay on a silicon wafer were in place. It's therefore no surprise that researchers have been working on just that - ultra small displays with extremely high pixel densities. I can't find any publicly available research papers, but here are a few links that contain abstracts from paid papers:

http://proceedings.spiedigitallibrary.o ... eid=813793
http://link.springer.com/article/10.100 ... 011-4498-y
http://ieeexplore.ieee.org/xpl/articleD ... er=5643334


Besides the obvious advantages of compact size and high pixel densities (think about the 5+MP imaging sensors in cellphones that are smaller than a fingernail to get the idea), they could be manufactured significantly cheaper than a LCD display ever could. Just as significant is the fact they could contain all interface circuitry on-die, meaning you could hook up a DisplayPort, HDMI, DVI etc. line DIRECTLY to the display chip. This reduces the costs of a device utilizing them even further than if you went the standard LCD route. With enough video signal bandwidth, refresh rates are unlikely to be limited by the display itself - thousands of Hz would be possible (but we would be happy with several hundred ).

I have little doubt that these will be the displays of choice for AR glasses and likely VR HMDs going forwards. I hope Oculus is paying close attention to the development of this tech, because this truly is a game changer.

[geekmaster]

QLED (quantum dot LED) displays look like a good bet for the near future too. There are a few links in this post:
http://www.mtbs3d.com/phpBB/viewtopic.p ... 25#p101425

[colocolo]

this display technology sounds very promising indeed, but somewhere i read Palmertech mentioning that these displays sizes wouldnt be appropriate for full 90 °immersion. cant remember exactly. has it do something with geometrical optics? i dont know...

[MSat]

QLED (quantum dot LED) displays look like a good bet for the near future too. There are a few links in this post:
http://www.mtbs3d.com/phpBB/viewtopic.p ... 25#p101425

I don't see QLED displays being manufactured cheaper than LED-on-silicon, to be honest. Even if they could make display itself cheaper, it would still need separate support circuitry. I also doubt that they would be able to get equivalent pixel density. Perhaps that's not the biggest deal for a VR HMD, but it would be extremely important for AR glasses.

this display technology sounds very promising indeed, but somewhere i read Palmertech mentioning that these displays sizes wouldnt be appropriate for full 90 °immersion. cant remember exactly. has it do something with geometrical optics? i dont know...

Considering that the DMD (the active imaging device) in a DLP, or the LCOS "display" in a TV or projector is ~1 inch diagonal in size, but the image you see is MUCH larger, I don't see many reasons why it shouldn't work. Even if it requires multi-element optics - since most of them would likely be small and compact, they shouldn't add any substantial weight or make the size too large.

[colocolo]

QLED (quantum dot LED) displays look like a good bet for the near future too. There are a few links in this post:
http://www.mtbs3d.com/phpBB/viewtopic.p ... 25#p101425

I don't see QLED displays being manufactured cheaper than LED-on-silicon, to be honest. Even if they could make display itself cheaper, it would still need separate support circuitry. I also doubt that they would be able to get equivalent pixel density. Perhaps that's not the biggest deal for a VR HMD, but it would be extremely important for AR glasses.

this display technology sounds very promising indeed, but somewhere i read Palmertech mentioning that these displays sizes wouldnt be appropriate for full 90 °immersion. cant remember exactly. has it do something with geometrical optics? i dont know...

Considering that the DMD (the active imaging device) in a DLP, or the LCOS "display" in a TV or projector is ~1 inch diagonal in size, but the image you see is MUCH larger, I don't see many reasons why it shouldn't work. Even if it requires multi-element optics - since most of them would likely be small and compact, they shouldn't add any substantial weight or make the size too large.

apropos substantial weight. do lenses that are very lightweighted exist? i mean it would be great if the oculus rift one day only weighs around 80 gramms, or 2-3 ounces. OLED displays could contribute to that and a very lightweighted case also. i am just imagining the possibilities
for fun
by the way silicon micro display is developing a 8k4k display.
http://www.siliconmicrodisplay.com/technology.html
which system could render side by side 8k4k or 64MPixels?

[geekmaster]

apropos substantial weight. do lenses that are very lightweighted exist?

To minimize lens weight, the lenses must either be narrow diameter or quite thin. Narrow diameter lenses must be very close to the eye to have an immersive FoV, which means either they are worn as contact lenses, or you have eyelashes brushing against them. Large diameter lenses can be kept thin and lightweight either by using fresnel lenses, or more exotic (and expensive) flat variable optical density lenses. There are also some interesting displays being built with integrated optical correction that eliminates the need for lenses altogether.

A good compromise in lens design for HMD use may be a hybrid lens with a small aspheric lens mounted in the center of a large fresnel lens. High quality fresnel lenses may be manufactured from materials with excellent optical properties, and may be a coated to reduce reflections and other optical distortion.

Also, for Rift-like HMDs that use SBS-Half video content, anamorphic lenses may be required to stretch the image half from portrait mode to landscape mode to fill more of the available horizontal FoV (like the shape of eyeglass lenses or movie theater screens).

The Rift Dev Kits have replaceable lens cups, so experimental aspheric anamorphic lenses may be used with them when they become available. A stacked pair of fresnel offset lenses may also be used to provide nonlinear anamorphic expansion, for inexpensive experimentation. An offset lens is cut from the outer portion of a larger aspheric or fresnel lens, and provides asymmetrical magnification proportional to its distance from the original lens center. A stacked pair of left and right offset lenses can stretch the image both left and right, eliminating any need for a center divider to prevent seeing a portion of the wrong image, and can provide additional peripheral FoV as well.

FYI, FoV = "Field-of-View".

[MSat]

I meant that with such a small display, most of the lenses in a multi-element design except for the eyepiece could be kept fairly small (in terms of diameter).

[geekmaster]

I meant that with such a small display, most of the lenses in a multi-element design except for the eyepiece could be kept fairly small (in terms of diameter).

Regardless of display size, because your eye can rotate in the eye socket, a wide FoV requires a lens that covers all of the available FoV (in every direction the eye can look).

A smaller display would require greater magnification from a THICKER lens than larger display would require for a given FoV. The 7-inch display can get by with a fairly low (5x) magnification,. Magnifying a tiny display for full FoV would require really high magnification, and such lenses may be more expensive or harder to find, and unless it is a compound lens may also have more chromatic aberration or other distortion. Thankfully, we can do a certain amount of correction for such optical defects in software pre-warp filters.

[druidsbane]

The good news is that I assume we can get the display at super-high resolution at any specific size we need since it is in silicon. My question is how far away from this tech are we and will it be much pricier to get it this way considering how expensive CPU's are based on die size?

[geekmaster]

The good news is that I assume we can get the display at super-high resolution at any specific size we need since it is in silicon. My question is how far away from this tech are we and will it be much pricier to get it this way considering how expensive CPU's are based on die size?

When we can get enough pixels in a small enough area, we can do holograms, and we will not need to deal with lenses. Now, a wall-sized holographic display would be ideal...

[MSat]

@geekmaster

Even if the eyepieces had to be a fair amount thicker than what the rift currently uses, it would likely be offset by substituting the weight of the LCD for the practically nonexistent weight of the microdisplay, as well as in the potential reduction in the size of the HMD housing. Also, since the weight would be brought closer to your face, it would reduce the amount of torque the HMD exerts on your head.


@druidsbane

I couldn't say for sure when we'll be seeing them on the market, but given that researchers apparently have working prototypes, and silicon substrate-based LEDs are expected to hit the market in around a year or so, I would say possible 2 to 3 years from now.

The maximum display size is limited by the size of the silicon wafer, but expecting a display in any physically large size is impractical from a cost perspective, as you have correctly pointed out in your CPU example. The major benefit of a CMOS-based microdisplay is that it has such high pixel densities, that it *doesn't* have to be large - making it very cheap to produce. Taking my earlier cellphone imaging sensor example - the sensor+optics module the iPhone 4S and the 5 (which are considered to be of high quality, and a resolution of 8MP) is estimated at $18! What would you do with 3264x2448 per eye?

[druidsbane]

MSat: you make some excellent points. So basically we'd use different lenses with really small displays attached at the ends... pretty interesting direction for this to take

[MSat]

I'm still a complete optics n00b, but if I'm not mistaking, a 2 lens array should work out to something like:

microdisp.jpg

The black rectangle on the left is the display, the blue lines represent the lenses, the red lines are the light rays, and the circle on the right, well, that would be an eyeball

The focal length will determine the thickness of the lens - the shorter the focal length, the thicker the lens. You can make it real compact if you use short focal lengths for the optics, but it's my understanding that by doing so, you increase the amount of abberation. So one way of keeping it compact while using thinner optics may be to place the display on the side of your face pointing forwards towards a mirror which then reflects the image back towards the eyepiece.

That said, I could be totally wrong about the optics, so take it with a grain of salt. I'm still learning

[geekmaster]

That diagram essentially shows a microscope, with a small LCD panel mounted on the microscope slide. A pair of these mounted together would look a lot like a pair of binoculars.

As an experiment, you can look through a pair of binoculars BACKWARDS, and it makes a great microscope, but with a very small field-of-view. I looked at my Nexus 7 display through reversed binoculars just now, and very tiny letters fill the display, clearly showing the R/G/B subpixels.

You would need a microscope that has a really wide FoV, to use a tiny display in an immersive HMD.

Perhaps a pair of eyeglass-mounted high-magnification jouler's loupe magnifiers would work. Or lenses similar to the ones recommended for the DIY Rift, but much higher magnification, and maybe another lens as shown in your diagram too, for these small displays.

[MSat]

In essence, it is indeed a simple compound microscope! I actually looked up "how a microscope works" after writing my post to see if it was a similar concept, and if I was on the right path

My understanding is that the FoV in such a setup is largely dependant on 1) the diameter of the eyepiece and its proximity to the eye, and 2) the focal length (or optical power) of the eyepiece - as in its ability to refract light towards the eye.

For binoculars, the big lens tends to have a long focal length, so even though it has a fairly large diameter, its optical power is not sufficient to provide a wide FoV because the light from the display exiting the eyepiece would be collimated (mostly perpendicular to the lens) instead of bending inwards towards the pupil.


Even if the objective lens had to have a short focal length (and thus, thick), its size is relative to the size of the display, so in reality it would be quite small and light weight anyway. As I mentioned, any significant weight due to the eyepieces would likely be offset by the reduction in the weight elsewhere in the system.

[Direlight]

When we can get enough pixels in a small enough area, we can do holograms, and we will not need to deal with lenses. Now, a wall-sized holographic display would be ideal...

Sony Hologram

[youtube]http://www.youtube.com/watch?v=0oWS0pBMR5E[/youtube]


Beacon 360
http://www.youtube.com/watch?feature=pl ... SnPgIHnMQQ

OLED will probably be the next new mainstream display type though. Lots of different types are in development though.