HMPD - has anyone built one of these?

Post by **ERP** » Wed Nov 24, 2010 10:55 am

Head mounted projection display.

They seem extremely simple to construct assuming you can get hold of some decent quality retro reflective material.
configurations seen http://vision.ai.uiuc.edu/publications/ ... ective.pdf and http://odalab.ucf.edu/Publications/2007 ... xpress.pdf.

I did some hand held experiments last night with a pico projector, a CD case and the strip of retroreflective material on my jacket, and it was enough for me to actually order some white retroreflective tape and a 50% beam splitter.
I suspect there will be some minimal distance at which the retro reflector fails to reject enough of the off axis light to provide anything aproaching a clean image but it's worth a play.

Just wondering if anyone else has experimented with this type of display?

Post by **tritosine5G** » Wed Nov 24, 2010 5:24 pm

This is very cool stuff indeed.

I'd wait for laser beam steering becoming more widespread than just one manufacturer. Currently the only commercial choice is microvision. Seems the tech is in its infancy, yet! However! Looks like texas instruments won't lock this tech like they did with DLP, seems like you can buy the micromirror:
http://focus.ti.com/docs/prod/folders/p ... 1000b.html" onclick="window.open(this.href);return false;
[youtube]http://www.youtube.com/watch?v=IlDG3PKT ... re=related[/youtube]

with laser, you can project onto any kind of curved surface without loosing focus.

Directly competing with LBS is syndiant:
http://www.syndiant.com/" onclick="window.open(this.href);return false; they can also use laser lightsource, but with a field sequential LCOS imager.
http://www.syndiant.com/tech_advantages.html" onclick="window.open(this.href);return false;

A compelling marketing point with LBS is that of “focus free” operation, but it should be noted that Syndiant’s VueG8 is also focus free when using laser illumination due to the optical properties of laser light. Syndiant’s design partners and customers have a choice of using either LEDs or lasers based on their cost and performance trade-offs.

Post by **cybereality** » Thu Nov 25, 2010 3:38 pm

This sounds very interesting. I have been wanting for a while to do a projection based HMD but I am uncomfortable spending the money without knowing if it will work or not. This idea sounds good on paper, but I am not sure I fully understand how it works. From what I am getting, the projector sends the source through the pass-through side of the beam-splitter, goes through a lens and then reflects off the retro-reflective material, back through the lens and then onto the reflective side of the beam-splitter and then to your eyes. Is this correct? I am not clear on what the lens is doing. I mean if light passes through one side and back, doesn't it reflect as if the lens wasn't there?.

Post by **tritosine5G** » Thu Nov 25, 2010 7:12 pm

Picoprojector biz is a battleground. Lets just wait .
I'd love to see "large" LBS pj around cuz I was a CRT maniac. LBS is capable of the same: perfect blacks. Not sure about syndiant, but with DLP, native contrast is better with a bigger device,so syndiant might have better resolution but worse contrast cuz its so small. IDK im no expert.

Here at Syndiant we are hard at work on our HD generation of products that we expect to sample this year and go into volume projection in 2011.

http://www.syndiant.com/images/PS%20Boy ... 2440_e.JPG" onclick="window.open(this.href);return false;

Post by **ERP** » Thu Nov 25, 2010 9:04 pm

Lens is just to converge the light onto the smaller retro reflective patch.
If you think about the beam path it would be much wider than the beam splitter at the point it hits the retro reflector, the lens just makes packaging easier.
At least that's how I read it.
It seems to me that there are better ways to package it.

I've read a couple of other papers, since and the limiting factor seems to be that retro reflectors aren't designed to be perfect, since they aren't designed for optical applications, so you get more of a cone coming back rather than a direct ray, which affects the image quality.

Laser projectors are interesting, but in this case unnecessary, unless you used a curved splitter to increase the FOV.
I have a pico projector sitting around, so I'll put a bench prototype together when my beam splitter turns up.

Post by **Okta** » Fri Dec 03, 2010 5:25 am

Can you guys explain this thing to me? I dont understand what its trying to achieve with all the reflecters? Isnt it just a see through hmd? Why not use 50/50 glass instead?

Post by **ERP** » Fri Dec 03, 2010 3:29 pm

It's all in the retro reflector.
The glass should be 50/50, but you're not seeing the image reflected off the glass, you're seeing the reflection from the retroreflector, passing back through the glass.
What you see is the image at the point it would be in focus on a screen.

Theoretically the distance the retroreflector is from the glass makes no difference to the image, since the light is simply returned down the same path towards the eye.
Practically retroreflective materials today aren't designed for optics, so there is scatter.

Maybe this picture helps

HMPD.JPG

Post by **Okta** » Fri Dec 03, 2010 6:09 pm

Still not understanding. How is that different to just viewing the glass with the projector shining on it like a HUD?

Post by **ERP** » Fri Dec 03, 2010 6:25 pm

Simply reflecting the projected llight from a 50% mirror would be equivalent to just looking down the beam of a projector at 1/2 brightness.
You can make this work, but the optics aren't simple.

If you're using some sort of difffuser instead of a mirror, then the diffuser acts as a screen, and you have the usual HMD focussing issues.

The difference here is you effectively have a screen, but because the screen is a retro reflector, it appears as if it exists at the focal depth of the projector.

At least that's how I read the papers.

Post by **cybereality** » Sat Dec 04, 2010 12:21 am

I thought about it really hard (and read some other whitepapers) and it finally makes sense to me. The thing that I was getting hung up on was thinking that the glass was being used as the surface of the image (meaning a projection screen). It is not. It is only being using to bounce the light around in an interesting way. When the projected image finally reaches your eye, it is just passing through the glass (well at least 50% of the light is, the other 50% is being discarded). The retro-reflector itself is, in some ways, the projection screen but since space is limited (with both the size of the retro-reflector and the entire light path), it is necessary to use the lens and the beam-splitter. Its important to understand the properties of the retro-reflector. It reflects light in a parallel, but opposite direction than it comes. This allows you use the lens to focus in on a very small area of retro-reflective material, but have it bounce back projected outwards (as if it were a projector itself). This is quite an amazing concept. I am still trying to wrap my head around the scale of this thing (as in the focal length of the projector combined with that of the lens and how big it would make this contraption). However it appears that high FOV may be possible, at least in the 60 degree range. However, I am no expert on optics (or what is available at affordable prices) so I may be a little off there. However this seems like it could be a relatively cheap DIY project. Here is a diagram I made to try to explain the idea:

I think the key advantage of the system is that neither the projector nor the retro-reflector need to be in your line-of-sight. All that will be in front of you is the half-silvered mirror, which will not be very distracting. So the HMPD could easily accommodate augmented-reality applications or similar heads-up-displays. It seems like it would be able to function with less complex optics than are necessary for traditional HMDs. So I think it would be easier to achieve high-FOV and still have a clear, undistorted virtual image. There might be other benefits too, but I am not exactly an expert when it comes to optics. But this is a very interesting method and I think it is just what I needed to make my dream of a projector-based HMD a viable idea. I've been spending too much money lately, but if I get a Christmas bonus, I think I know what I'm buying.

Post by **Okta** » Sun Dec 05, 2010 4:44 am

So going by your picture cyber, you can set this up like an HMD where the RR becomes the screen magnified by a lens in front of it?

I think i just got it, this will be like using a single monitor planar setup for each eye. The RR is to turn the projector into a monitor that will be seen as though you are looking through the 50/50 glass.

It really didnt make sense until i thought about it in terms a a planar setup.

This opens some doors for different size screens and projectors that would be too large side by side i think...

Post by **cybereality** » Sun Dec 05, 2010 12:18 pm

Well this is a little different than a planar setup, although they both use a beam-splitter. With the HMPD, the light that reaches your eye is coming through the 50% of the glass that is see-through. The other 50%, the reflection, is being thrown away. This is in contrast to a planar setup, where you still only get 50% of the light, however, the light that reaches your eye is bouncing off the 50% reflective part of the glass. The light that passes through is being discarded. The only thing that is the same about the setups is that you are ultimately viewing a reflection of the original source, and not directly viewing it. If it helps to think of this in terms of planar, thats cool, but they are a little different.

In some ways you can think of this HMPD setup as similar to simply shining a projector directly unto your retina. In fact, I am still not sure how the setup shown in the diagram fundamentally differs from projecting directly into your eye. Yes, the light is bouncing around somewhat, but its just being reflected. There are not any diffuse elements (assuming you had a perfect retro-reflector, like a corner mirror. Retro-reflective paint or fabric will likely have diffuse properties). I think this may cause some safety concerns, especially with laser projectors. From what I read, everyone is saying that looking directly into a projector beam can be damaging to your eye (like looking at the sun) and lasers have always been said to be dangerous (although the type used on pico-projectors may be within safe limits. I don't know). So I will have to read more about this before I become a guinea-pig. If I lost my eyes, well, I just don't know what I would do (besides trying to wet-wire a camera to my brain...). Ah, the dangers of living on the cutting-edge.

Post by **PalmerTech** » Sun Dec 05, 2010 5:09 pm

I have two sets of optics just like this, one semi-occlusive, the other fully so. They come from the really old i-Glasses.

The retroreflector is curved, and provides high magnification of the image.

I wish I had a working pico, I am guessing that if the focal length were set to where the LCD used to be, it would work very well.

Post by **cybereality** » Sun Dec 05, 2010 7:26 pm

I know I said I was going to save my money, but now I am looking at some pico projectors. I found a lot of stuff in the $100-$200 range, all 640x480 mostly with RCA input only. The ShowWx seems to be the holy-grail at the moment for $300: laser projection, perfect focus at all depths, good brightness. However I think my interest has been piqued by the EyeClops Mini Projector, which is only $40 with free shipping: http://www.amazon.com/Eyeclops-04959-Ey ... B0024I8H42" onclick="window.open(this.href);return false; . I could easily spend more than that on a few beers at a bar. So the price is tempting. Its only 320 x 240, and the quality is low, but that is to be expected from a "toy". Do you guys think this projector would be able to work for an HMPD? Obviously it is just for a proof-of-concept. If the idea works then I could drop a few hundred on some better hardware. I guess for $40 it almost doesn't even matter, I'm sure I could find some use for it. What do you guys think?

I might still wait a little bit since I have a lot of research to do regarding the optical lens and also finding a good supplier for a half-silvered mirror (beam-splitter). I did find this site, and they have cheap 4" x 4" samples ($3) although only at 60/40 transparent: http://www.telepromptermirrors.com/tele ... sglass.htm" onclick="window.open(this.href);return false; . Might be sufficient for testing, though. However you can make a custom order for 50/50 but who knows how much that would cost. Can't seem to find any other decent suppliers that don't want an arm and a leg. In terms of optics, I found EdmondOptics, which seems to have a wide selection and good prices: http://www.edmundoptics.com/onlinecatal ... egoryid=11" onclick="window.open(this.href);return false; . However I have no idea what any of those numbers mean, so I don't know what I am looking at. However with that kind of selection, I'm sure something will fit our needs.

Post by **Okta** » Thu Dec 09, 2010 4:16 am

Cyber, you should be able to get a vga projector with xga res for $100.
I wonder if this http://cgi.ebay.com.au/ws/eBayISAPI.dll ... K:MEWAX:IT" onclick="window.open(this.href);return false; would be ok for testing?

Post by **cybereality** » Thu Dec 09, 2010 10:12 pm

Okta wrote:Cyber, you should be able to get a vga projector with xga res for $100.
I wonder if this http://cgi.ebay.com.au/ws/eBayISAPI.dll ... K:MEWAX:IT" onclick="window.open(this.href);return false; would be ok for testing?

I did see some units in the $100 range, but they were all 640x480 (or less). The higher res ones seemed to be more money, although I did not look for very long.

Not sure if that reflector would work, but you can't really go wrong at that price. I did read that type of material (like the RR tape, stripes on jackets, etc.) are designed to be slightly diffuse and not a perfect retro reflection. I think they said the reflection angle would be +-3 degrees. They do this since the main application is for driving safety, and a perfect retro reflector wouldn't work. If it were perfect the beam of light would only reflect directly into the car's headlights and not be visible by the driver. So there must be some leeway. This of course is not what we want for precise optical applications. However for a quick proof-of-concept, it might suffice.

Post by **Okta** » Fri Dec 10, 2010 5:29 am

The xga i refer to is just upscaled from vga of course but still handy for fitting menus on screen. I agree those reflectors wouldnt be much good, would just give a colored smudge.

Further thinking.. Do we really need an RR for this to work? It should work with a normal piece of projector material as we are just looking at the reflection of that from the glass. The killer will be the light loss which is maybe why the RR is needed to boost the light reflection. If we use this as a HMD and not a see through AR device it will be enclosed so the light loss should not matter as much.

Got to get of my butt and do some testing...

Post by **cybereality** » Fri Dec 10, 2010 9:55 pm

Okta wrote: Further thinking.. Do we really need an RR for this to work?

Yeah, we do. If we use a regular mirror, and focus the light into a small area (like in my diagram) then it will reflect getting smaller. I guess if you set it up right this could work (as it would eventually reverse the image and get bigger again) but the image would be backwards. Using a projector screen material would also somewhat work, but it would be no different then just strapping the projector to a helmet and looking at the wall. Although I would be interested to know what method they are using in this video:

[youtube]http://www.youtube.com/watch?v=AoO0kJavV80[/youtube]

Seems like they are just projecting directly on a piece of glass using a laser projector. Does that work? Why doesn't the image just go through the glass?

Post by **PalmerTech** » Sat Dec 11, 2010 11:44 am

I ordered a ShowWX yesterday, I am going to be running a barrage of tests on it. I will let you know the results.

I might even try retinal projection. Turn it to the lowest brightness setting, and apparently the new firmware allows for individual RGB level controls. Going to turn all the levels to minimum, then use my LPM to measure the combined output of the projection. If it is under 5mw on a point, then I will be shooting it into my eye.

Post by **ERP** » Sat Dec 11, 2010 7:43 pm

Do what I do, when trying retinal projection, put 2 sheets of polarized material in front of the lens, change the relative angles to adjust the brightness.
My experience is that the exit pupil in minuscule.

Post by **ERP** » Sat Dec 11, 2010 7:48 pm

Seems like they are just projecting directly on a piece of glass using a laser projector. Does that work? Why doesn't the image just go through the glass?

Yes it'll work, with a rather large loss of brightness.
All glass is a specular reflector at none 0 angles of incidence, a lot of light will go straight through though.

Post by **cybereality** » Sat Dec 11, 2010 8:34 pm

PalmerTech wrote:If it is under 5mw on a point, then I will be shooting it into my eye.

Good luck man. I sincerely hope you know what you're doing.

Post by **3DHMDGuy** » Mon Dec 13, 2010 12:07 am

Not sure if this will help but this projector is about $300 can do Resolution: 1024 × 768

http://www.walsoon-tech.com/en/index.ph ... &Itemid=82" onclick="window.open(this.href);return false;

If anyone get round to making a high res, high FOV HMD. Have got cash waiting for 1000 units. I'm serious. Price needs to be sub $1000. PM or write here for contact.

Post by **ERP** » Mon Dec 13, 2010 12:04 pm

cybereality wrote:
Okta wrote:Cyber, you should be able to get a vga projector with xga res for $100.
I wonder if this http://cgi.ebay.com.au/ws/eBayISAPI.dll ... K:MEWAX:IT" onclick="window.open(this.href);return false; would be ok for testing?
I did see some units in the $100 range, but they were all 640x480 (or less). The higher res ones seemed to be more money, although I did not look for very long.

Not sure if that reflector would work, but you can't really go wrong at that price. I did read that type of material (like the RR tape, stripes on jackets, etc.) are designed to be slightly diffuse and not a perfect retro reflection. I think they said the reflection angle would be +-3 degrees. They do this since the main application is for driving safety, and a perfect retro reflector wouldn't work. If it were perfect the beam of light would only reflect directly into the car's headlights and not be visible by the driver. So there must be some leeway. This of course is not what we want for precise optical applications. However for a quick proof-of-concept, it might suffice.

I've tried white/silver retroreflectors and the issue is that the standard diffuse reflection washes out the retro reflection.
It'd probably work better if the retro-reflector we're further away, since that would reduce the ambient reflection, but shouldn't affect the retro-reflection.
I've ordered some black retro reflective material to see if it's any better.

Post by **ERP** » Fri Dec 17, 2010 2:50 pm

I have my black retro reflector, and now I think the white/silver ones are probably fine.

My understandong of the optics was somewhat flawed, I need to sit down and draw some to really understand what's happening.
I need to build a more stable bench prototype, but doing some hand held tests last night, as far as I can tell the following is the case, the projector has to focus on the retro reflector (practically this isn't much of an issue). But the image seems to appear in focus regardless of the focal point of the eye, if you don't use a retro reflector, you have to focus at the effective depth of the screen.

This is not what I expected, as I said I need to do some more testing, and I need to understand optically what's happenning.

I should note that what I'm observing with tests is still promising, it's just not what I expected.

Post by **cadcoke5** » Fri Dec 17, 2010 9:09 pm

I have played with retro-reflectors, but not in relation to VR. Rather, I was working on a project to put a laser beacon on a robot, and then when it hit a retro-reflective target, the reflected light would be sensed by a photo transistor. This would be used to calculate an angle between the targets, and then used this data to determine location.

Sorry for the long 0ff-topic explanation, but the point was to say that when I used a single point laser, the reflective
light would trigger the sensor very on-off-on-off, etc. This was because there was a grain to the reflectors. As the laser hit one spot, it would be reflected strongly, but then slightly to one side, the reflection was aimed off a slightly different direction, and not sensed. I solved this by putting a beam spreader into in to generate a fan of laser light. But, this created some of its own challenges.

I used some retro-reflective tape I purchased at Walmart, designed to be applied to things like trailers. What exactly was the source of your retro-reflective material?

Joe Dunfee

Post by **cybereality** » Fri Dec 17, 2010 9:17 pm

I think you can also use a corner-reflector for the same purpose as the retro-reflective material. In this case you can just use standard 100% reflective mirrors so they should be optically accurate.

Post by **ERP** » Sat Dec 18, 2010 2:40 pm

It's definitely acting as a retro reflector, if I put a Lens between the mirror and the screen, I can see it distort the edge of screen, but not the image. which us what I would expect.
It's just the way I read the original diagram in the papers is incorrect, I need to think about what I'm seeing.
The grain is clearly visible on all of the retro reflectors I've tried. A corner reflector is an interesting idea.
One of the bigger packaging issues is going to be the short throw of the projector, you really want optics to give you a longer throw, or the projector has to be right on top of the mirror.

Post by **cadcoke5** » Sat Dec 18, 2010 3:13 pm

Cyberreality, in my case I had one more complication that I didn't mention. I was planning on doing barcode with the reflective tape. So, a corner reflective would not work in that case. I think the tape is, in essence, a bunch of tiny little corner reflectors. Some of the brands I have seen are coarser grained that others.

Joe Dunfee

Post by **ERP** » Sat Dec 18, 2010 8:01 pm

OK after further messing around, I think it is working the way I originally expected.
The issue is the diffuse reflection is sitting on top of the retro reflection, and the projector is so bright, that it's hard to differentiate.

Post by **Okta** » Sat Dec 25, 2010 10:49 pm

I just did a simple experiment to get a feel for this idea as seen the the picture below.
At the top is an lcd (i used my phone) shining down through the clear cd lid, it then reflects off the mirror at the bottom back up onto the eye side of the cd case which is viewed as a floating screen through the cd case and is the correct orientation (not reversed).
This in itself is like looking at the phone normally except that it is not as bright. The only geek use is that this is now an Augmented Reality device as the screen is superimposed over your normal view. The down side is you cant see it in daylight because of the brightness. Im considering options to use one for each eye but one up, one down to make size for bigger displays that would normally clash if side by side.

AR.JPG

I then added a magnifying glass on top of the mirror and it did add some decent size increase to the image as expected. The result with my 2.5 inch phone lcd was a deal bigger than the vr920 view.

AR2.JPG

I then tried replacing the lcd with a pico projector. This didnt work for a few reasons. Firstly it projected a reflection directly from the cd case onto the wall i was looking at (in reverse). This might not be so bad for AR as long as you dont get to close to something or someone. Secondly it just reflected tons of light straight back into my eye like looking straight into the projector. This has me still wondering what would be the result with an RR instead of a mirror.

AR3.JPG

Now my understanding of the full concept is that the only reason the RR or mirror is used is to correct the orientation of the image so it isnt reversed. I am still not sure that an RR is needed at all rather than just a mirror and the correct glass/filter/lens combo.

Post by **cybereality** » Sat Dec 25, 2010 11:09 pm

Hmm, interesting. Looks like the method still needs some work (or rather, maybe the 50/50 mirror and RR are more essential than you think). Although, if you are getting a bigger image than the VR920 even with such a make-shift setup, we are at least getting somewhere. I'd like to hear how this progresses.

Post by **Okta** » Sun Dec 26, 2010 2:31 am

Epiphany!!!!

Sorry if this is getting too off topic (should i make another thread?) but after dwelling on my last post i have had this idea.

Note i haven't done a mockup of this yet but its using the same idea as before but with added magic (yay magic!).
Firstly instead of up/down its side by side. Its basically a method to use 2 large lcd's (4.5" maybe) with twin mirror method because of side by side size constraint. We know 2 mirrors are required each side to move the image AND correct the reversal. But this uses instead the 50/50 glass as one of the mirrors. The difference here is that because of placement you can see the lcd's through the glass but you dont want to. This is what the polarized filters in front of each eye are for (not sure if more will be needed). Hopefully they will block the direct through glass view and only allow the 'virtual' reflected view.

Playing with sizes/positions this may give us a healthy FOV reasonable res hmd.
I feel all excited about this like i should patent it lol. This could be a big ugly but useful rig for under 1k.
What do you think?

Post by **cybereality** » Sun Dec 26, 2010 10:14 am

Ok, that looks like a good idea. However, what is the point of having the lens right on top of the mirror. Wouldn't it work better it it was right in front of the LCD or right in front of your eye (ie in a position that light only passes through once). If its right on the mirror, the light would pass through and then bounce right back effectively negating the purpose of the lens. Right?

Post by **ERP** » Sun Dec 26, 2010 12:56 pm

I think you'll have to move the displays out of the line of sight , because the two images won't line up.

Post by **cybereality** » Sun Dec 26, 2010 7:50 pm

ERP wrote:I think you'll have to move the displays out of the line of sight , because the two images won't line up.

True. I didn't even notice that. You would be seeing some *serious* ghosting if the LCD was in your line-of-sight.

Post by **Okta** » Mon Dec 27, 2010 2:57 am

cybereality wrote:
ERP wrote:I think you'll have to move the displays out of the line of sight , because the two images won't line up.
True. I didn't even notice that. You would be seeing some *serious* ghosting if the LCD was in your line-of-sight.

Thats what the filters in front of the eyes block out, the actual lcd screen. If it works out that it also block the virtual screen a second filter will be needed over the mirror maybe.

This setup is not for AR, this will be enclosed.

Post by **ERP** » Mon Dec 27, 2010 12:27 pm

Interesting, missed the filters, then yes it ought to work although I think you'll need some way to additionally rotate the polarization of the final image.
You have 2 bounces off mirrors which will leave you where you started, I think...

Post by **Okta** » Sat Nov 05, 2011 5:33 am

John: My previous test with a projector shining onto the beam splitter just made the light reflect into my eyes like a torch without any image. What did work was shining the projector through a beam splitter onto a screen surface, that image is then seen as a virtual screen through the beam splitter. This would certainly be the easiest way to use pico's in a DIY HMD i think although it is difficult and bulky still to get a large FOV.

Get your phone and a clear cd cover or glass sheet in a dark room to experiment.

Post by **Okta** » Tue Nov 08, 2011 7:17 am

I hadn't really thought about that as i intended to make a closed system anyway. I dont see how that could be done? At the end of the day it is the lens on the eye that focuses a set distance for the retina so i cant see how it could be done?

Oh, i also used a mirror and had the cd facing away in my test as per my pictures before, that make the screen the correct orientation.