I found some interesting patents for super wide fov.
The Hmd of Nagahara has a vignetting problem but there are other patents also with elliptic mirrors. Prehaps devices without vignetting.

example: http://www.freepatentsonline.com/6522474-0-large.jpg

The optics has to be aligned with the eye surface to avoid vignette.

This is my first construction trial of a elliptic mandrel.

I'm very curious about this project. I assume this is to grind your own mirror?
I think that the finishing of this with such a tight radius will be difficult without deforming the shape.

The design you show is common, the old i-Glasses units look exactly like that. I don't think it helps achieve wide FOV, though.

It is possible to make deep-curved mirrors from plastic.
The only problematical case is the mandrel surface, it has to be smooth and polished.
Some spherical mirrors are made from silvered ABS or PMMA plastics.
The mandrel is used for plastic vacuum forms (double layer). the first plastic layer gives a smooth surface to my mandrel (would not be removed), the second one is removed for silvering
between layer 1 and 2 a special layer is used against sticking problem between surface 1 and 2.

update.
Turned ellips (before lacquer work).

@ 3dpmaster: this is really interesting!! Please keep posting your progress, as i'm very interested myself in the possibility of using mirrors rather than lenses to achieve high FOV.

I'm not even sure exactly what I am looking at, but I'm impressed.

Thanks for comments,

I try to make a super wide field of view HMD based on the device of Hajime Nagahara et.al. (180 degrees!!!)
My first project seemed to came to a dead end. Wide FOV eyepieces are too heavy and they give a maximum fov of 100°.
Even if the eyepieces are rotated away from the parallax alignment, they can reach an angle of 140° together (some hmd's like SEOS(TM) only to 120°).

I will check the prototype HMD of Mr. Nagahara's by making the same HMD by myself.
There is enough documentation (for me) to make the same catadioptric HMD.
This patent for example: http://www.cs.wustl.edu/~pless/omnivisF ... gahara.pdf
By hacking the Sony HMZ-T1, I'll get two high quality 720p oled microdisplays for my HMD. furthermore, I need one 360 deg lens for the camera, two hyperbolic mirrors for the HMD, two ellips mirrors, etc. .
I know, it is very hard to make diy custom mirrors, but it is possible!
There are different ways to get a deep curved mirror (ellipsoidal or paraboloidal). I try to contact company's specialized in electroplating.
If my project is on the right track, I'll post some pics.
I'm curious to know how the catadioptric HMD of Mr Hajime Nagahara works.

What kind of software do you plan on using to drive this? There are no 3D drivers out there that support adjustable stereo overlap, as far as I know. Will this be for custom applications?

Reply to PalmerTech

The 3D overlap is not important because the cylindrical properties of the image are merged with the 3D overlap angle. Only the interpupillary distance (IPD) is important for this wide FOV setup. The image plane is not flat but spherical (Dome). The eyes are always looking in front of the 2 vertical 'domes'. I have to check these spherical geometries.

I do not have any experience with 3D hardware yet, I'm still working with diapositives. In the near future I will buy and hack a Sony HMD. If there are HDMI inputs in the HMZ T1, I can test the omnidirectional video's on the 180deg FOV modified HMD.
The HMD will be custom but if I can, also for gaming etc. .There is an image problem because it has to be like this: http://www.cs.columbia.edu/CAVE/project ... _donut.jpg
This is important if you don't want to struggle with sophisticated algorithm software. A simple fisheye lens cannot be used, I have to use an omnidirectional lens with a hyperbolic geometry.

I'm sure, it is possible change the 3D overlap with software. The pair of "the two virtual viewpoints" can be spread from the parallax direction to determine the 3D overlap in the HMD.
In this case, the 3D overlap would be a software related problem.

I found an article about a huge HMD from toshiba.
article:
"A Japanese electronics maker Toshiba Corp. employee demonstrates a full-faced prototype headgear that enables her to get a close to 360-degree image on a dome-shaped screen while the same image is displayed on a monitor screen at Toshiba Corporate Research and Development Center in Kawasaki, west of Tokyo, Oct. 23, 2006. Toshiba, which plans to market the gadget within 2 to 3 years, aiming at attracting people such as virtual reality gamers, said the wearers can get 160 degrees in a horizontal plane and 120 degrees in a vertical plane from any angles their heads turn. (Xinhua/AFP Photo)".

Nice try but:

The screen seems tooo close to the observer and the HMD is SOOOO BIG!!

I love that HMD. That thing is the ultimate.

LOL. That is ridiculous!!!!!

180 deg fov Project update, test mirror for 180 degrees fov (ellipsoidal).
The image overall is impressive but the optical precision and focus coming from the lens should be better.
Focal distance = 70 mm
FOV = 180 degrees per one eye.
I have to make one mirror per eye and look for a better technique to make ONE shaped mirror.

I'll send new images as soon as possible.

This looks promising.

Looking very interesting! Keep us updated, this looks pretty phenomenal.

Did you see the pic of the lady when she got out of that washer on her head, It wasn't pretty, I think she also has a lawsuit still pending!

Damn... can't see the pics. I guess the links for pics are still being worked on with the new upgrade.

Yeah, annoying. I hope they fix that soon...

New update.
New trial with metal (tin alloy or Sn) for smooth mirror surface.
The mirror piece is casted in plaster (additional metal casting) then polished. It was very easy to cut off the flat bottom disc. The metal is very soft. (softer than aluminum), easy to polish and grind. A wooden ellips and stitched sandpapers are used for maintain the optical surface.

To be continued .....

Cool.

This project is interesting to me but so far beyond my expertise. I have nothing to add but words of encouragement. Keep up the good work...

Yeah. This is looking really awesome.

Being that your curved mirror is after the glass mirror the distance to the eye reduces the possible fov, maybe a large lense can increase this again? Also, shouldn't your curved mirror actually be white projector screen surface? In fact, i dont see the benefit of the curve at all looking at that design link....

The curve greatly magnifies the image, that is why it does not need traditional lenses. A projection surface only works with a collimated light source, such as a projection, and then you still need something to modify the focal length.

For a very wide fov, I skip the planar mirror, also H. Naghara did it in his device (illustration 1 & 2)
In this way, the field of view wouldn't be reduced. The curved mirror has to be as close as possible to the observer. If not, the rays cannot reach the second (hyperbolic) mirror. All the rays are reflected and guided by both mirrors (ellipsoidal and hyperbolic mirror). The hyperbolic mirror is not made yet. Therefore I need a hyperbolic mandrel.
If I have the hyperbolic mirror, everything will be mounted together with a standard lens or eyepiece (illustration 2).
Now I try to make a better curved mirror without gaps and holes. A lot of work

The pictures do a wonderful job of explaining the theory. Just a couple of questions though:
1. are you projecting from the side, bottom, or top?
2. Do you need the Planar mirror at all? If you are projecting from the sides, wouldn't it be easier to just project directly onto the hyperbolic mirror? Or is the focal length of the projection lens not acceptable for this?
3. I've been wondering if it would actually be possible to use slotted mirrors with the slots aligned and project through the front of the mirrors?

Thank you for posting this thread here, it's fantastic to see the experimentation and results. I just wonder why there aren't more commercial manufacturers creating products like this. I think if you can get this working acceptably, it would be a great one for kickstarter to get it to actual production. Obviously depending on your own desires.

Thanks for comments.
The pics are from a patent. If I don't use it for commercial use, it's ok.
There are several fov researchers like Kaiser, Leepvr, Sencics and Seos but their devices are too expensive for large scaled production (rated price = upto 200 000$ or more).

The (ellips) technology from the patent is one of the simplest way to get an ultra wide fov.
answers to 1, 2, 3.
1. The image is projected from the bottom
Projecting from the sizes would be difficult because the distortion from the left and the right image would be mirrored one to the other instead of duplication.
2. The planar mirror is used for avoiding the projecting interruption from the observers face.
3. There are reflection problems with see through mirrors in this device.

I will continue my experiments with everything from Diy shops, picking up rubbish and trash. Everything I can get for the ultimate HMD.

http://www.siliconmicrodisplay.com/st1080-features.html




Wait a minute, is this 1080p ????
180grams??

When will it be on the market?

I will use it with my mirrors.
This is the most awsome thing I ever seen (for dissassembling and make the ultimate hmd)

Pre-orders are available now. $699 if ordered before march 21 then $799. Shipping is set for May for the US

Where have you been? It's been the topic of discussion for the last month.

You found it just in time to get the pre-order discount.

Too happy too early...
500 usd or 1080p, let me guess....
Lcos is too delicate to handle with. Three screens are packed in a beamsplitter-box full of controllers, difficult to measure the focal length from it.
I think I will choose oled displays, more stable compared to Lcos. Oled is nothing more than a flat screen. A bit confused by the lcos technology. see pictures below
Cinemizer, 720P is not bad at all + shipping fees from Germany to Belgium......

Yeah, but the Cinemizer is NOT 720P.

I do not understand.
I mean there are:
1080p TV screens,
1080p monitors,
1080p camcorders,
But there is only one 1080p HMD on the market?
Why aren't they just follow the technology of today?

Does the Sony have the real 720p HMD? Hopefully, else I have to wait again and again.

Yes, Sony has a 100% native 720P resolution. It looks great.

Thanks!
I 'll buy it in the near future.

Currently, the accepted input format for HMD's is HDMI 1.4a. This does not have the bandwidth capability to support more than 1080p@30Hz which is not that pleasant. This means for the best experience, halving the resolution would allow you to double the framerate within the bandwidth constraints. Because of this, standard output is 720p@60hz which, when scaled up on a 1080p panel does not look very nice (without lots of AA, but even then...). Thus, in order to produce the best picture quality with the current format, a 720p panel is the logical choice.

Dual DVI or display port are capable of 1080p@60hz but I think (please correct me if I am wrong) there have only been specific model graphics cards that support 3D with dual DVI and as for DP... anyone know?

Not sure if I'm quite understanding correctly... Dual link DVI supports 1920x1080 @ 120hz, from most any DVI video card (I think).

Certainly my GTX460 and my GTX580, can both drive my Alienware monitor @ 1920x1080 @ 120hz.

I'm not saying 1080p@60hz is the maximum capability, I'm saying 1080p@60hz is a minimum for comfort. Dual DVI output is possible with most cards, but can throw up errors when mixing in different 3D formats (and not many of the less expensive consumer HMD's use Dual DVI).

The new mirrors are not ready yet. So I post some old memories from April 2003.
A diy polishing machine able to make any shape of lenses, meniscus, half spheric, two sized, ....
All the lenses were made from casting resin. Similar to the Leep optics, I tried to make a super wide fov viewmaster with 6x6 prints. The primary lens on the polishing machine is a very special one. It has a half spheric hollow side and a spheric size over the middle.
The lens is shaped in the form of the observer's eye socket. Result= Full field of view.

However, the assembly is too heavy and the 'rainbow' abberation at the edges is so annoying.
The process was very exciting and instructive, now I'm able to make a small lens with a drill press and sandpaper sticked on a cylindrical shape. No chisel, no lathe, no milling machine, no cnc, .,.,...

Cool.