Review of the Wide5

[PalmerTech]

I am sure some of you guys have seen the Wide5 (http://www.fakespacelabs.com/Wide5.html" onclick="window.open(this.href);return false;) before, but there is not really that much info out there on it. This review completely ignores the context of my using the Wide5, since that is for a different thread, but here is a review of the actual hardware itself:

Weight/Comfort: The Wide5 is primarily constructed out of carbon fiber, with a plastic frame to support the carbon fiber sheets. It weighs less than a kilogram, and while I know an entire kilogram sounds can sound like a lot, I have to say that this is one of the most comfortable HMDs I have ever worn! The weight is very evenly distributed, with no single part weighing a significant amount. The effect could be described as similar to having a large box full of foam peanuts, and a smaller box full of metal screws. While they may weigh similar amounts, one of them feels much lighter, and is easier to spin around.

The front of the HMD is very lightweight, and because of that, does not really need a counter weight. The frame does, however, have a carbon fiber rod that extends out the back and carries the cable. This balancing act means that all of the weight is centered on the top of your head, where you feel it the least.

All of the straps are lightly cushioned with breathable fabric covered foam, similar to what you see in bicycle helmets. The ratchets allow you to adjust it very quickly and easily to fit any head size. The lenses of the unit are a comfortable distance away from your eyes, and can be used with even very large glasses.

Driving the unit is a backpack driver box that controls the Wide5. It takes a single custom DVI input, and the guts convert it into the distorted image that the Wide5 HMD unit needs.


Optics/Resolution: I should not talk about the hard tech specs, unfortunately. I can, however, talk about perception of them as by an end user! The lenses are of fantastic quality, with no chromatic abberation visible to the eye. The lenses are focused to infinity, much like a fresnel monitor setup, and the focus is clear all the way to the edge. In terms of distortion, there is some, but in a good way: In a fashion similar to LEEP optics, there is a higher pixel density in the center than the edges, but even on the very edges of the optics, the image is clear. The exit pupil is enormous, no need to adjust the IPD or anything like that. The design of the panels used in the Wide5 makes it nearly impossible to see the pixel grid, you have to look very closely to notice it.

Field of view: This is where the Wide5 really shines. It has a horizontal FOV of over 150 degrees, and a vertical FOV of about 130 (Completely covers vertical FOV). The effect of a FOV this large will be covered in another thread (Will edit a link in here when I write it) is hard to even convey. The image ENGULFS you! Most HMDs try to emulate a large television screen, at distances of no further than 12 feet. Has anyone here used a fresnel setup, and felt the effects of the immersion that being able to focus and converge into the vast distance gives you? When you look off a bridge, across a river and on to the snowcapped peaks in the distance, you feel like you are THERE! Your eyes focus as they would really looking at something that far away, and the convergence is near parallel.

I will cover this more later, since it is more related to the software side, but read this article on orthostereo if you have not: http://www.leepvr.com/37spie1990.html" onclick="window.open(this.href);return false;

Orthostereo is, basically, the idea that what you see in the HMD perfectly matches what you would see in real life. Angles, rotation, etc. You can sort of get an idea of it by feeding a low FOV HMD a rendered image of 35 degrees or so with head tracking, but that is far too narrow to feel any immersion. When you have over 150 degrees to play with, there is no such limitation! Objects look exactly the same size they would in real life; Walk up to an armored humvee, and it has the same imposing presence it would in real life. Another virtual body waves their gun in your face, and you become genuinely concerned for the safety of your eyes. There is no concept of "screen depth" or "out of screen effects", it is just THERE, as it would be in real life.

Price: You essentially cannot buy one. At one point, several years ago I believe, a member here got a price of $35,000 quoted, but even if you had that money in hand, you would not be able to buy one. At the moment (As far as I know), there are no spare units just lying around waiting to be sold, so it would have to be a custom order.

I think that pretty much covers it! If you have any questions, please ask them, I will answer if I can.

Oh! Pictures! This fake head is a bit small, so if you are thinking "Hey, the lenses are too high!", keep in mind that it is an itty bitty foam head.













Wow, especially that last pic shows how small the foam head is. See how far out the strap sticks? The Wide5 seems a lot smaller on a real person's head, but nobody was free to help me take pictures of mine at the moment.

Archive note: If these image links break, even if years in the future, please PM or email me so I can fix them.

[WiredEarp]

150 degree FOV? I want. Badly.
There is no real comparision to the immersion you get with 90 degreee plus FOV, compared to the crappy FOV we are getting with our consumer level HMDs ;-(

[PalmerTech]

Remember how I was working on DIY HMDs for a long time? Well, those helped me get my new job, I will be posting some pics later tonight.

And yes, high FOV is absolutely critical. Keep in mind that the apparent screen size of the Wide5 is more than twice as large as a 90 degree FOV, absolutely stunning.

[Freke1]

Interesting to see. Now why is 2 small screens and 2 small fresnel lenses and some plastic so expensive?

[Okta]

OMG dude, that is yours?

Those lenses are amazing, just as i expected they need to hug the nose to get a real FOV.
But where are the specs? What trick does it use to redirect the eyes to the wide screen setting?

I am so jealous...

And what is your new job? Are you going to design us our dream consumer HMD?

[ERP]

I'd be interested in how the optics are configured.

[PalmerTech]

@Freke1: Small, high resolution panels are expensive, and the lenses are not fresnels. They are custom made glass optics, of a very novel design. No way to make those out of off the shelf parts, and only several sets exist.

@Okta: It is "mine" in the sense that I am allowed to use it at my job. Beyond that, it belongs to my employer. And if I understand your question about what trick they use, the lenses are slightly diverged, so the edges are out of your vision. As for my job, I am working at a research facility that builds a lot of really cool stuff, including high end combat simulators.

@ERP: Do you mean how they are configured internally? It is actually a very simple setup. Two small screens side by side, and a pair of lenses in front of them. I would elaborate on the lenses, but their design is one of the things I specifically cannot talk much about.

[Synexious]

Price: You essentially cannot buy one. At one point, several years ago I believe, a member here got a price of $35,000 quoted, but even if you had that money in hand, you would not be able to buy one.

What nonsense is this? Why can one not have something if they can pay for it? And why are these high-FOV HMDs constantly kept in niche applications? Sony has shown that mass production can significantly drive prices down. Is there some vested interest in preventing these HMDs from reaching the mainstream market? Don't the companies making these HMDs realize they could actually make more money with mass production? Don't they realize the huge benefit to humanity this technology could be? Fully immersive HMDs are useful for more than gaming - when the public finally gets their hands on HMDs that approximate actual physical presence, the need to travel will be drastically reduced, and people will buy fewer things. That will revolutionize economics and environmental preservation. High-quality VR is the ultimate cheat - it enables a minimalist lifestyle that is nonetheless extravagant. Because objects are built of light rather than photons, in the metaverse you alkyd get something for nothing. I'm supremely frustrated that ewe have all this amazing technology (not just HMDs - motion and haptic devices, too) which could allow for very convincing cyberpresence, yet no one's developing it for the mass market. If only people could see what they're missing.

[PalmerTech]

Keep in mind I could tell people where I work, but if I do, our employee handbook is very clear that I cannot act as a spokesperson or representative. For that reason, it is easier if I just keep information about my job to a bare minimum. Right now, the most I can say is that I work for a military research and development facility that deals with virtual reality, augmented reality, and simulation technology, among other things.

As for the rest... Realize that most of these companies essentially work for the military. Nvis, Sensics, even Emagin, their largest market is military, with a little bit of commercial/entertainment/university sales on the side.

Read this wikipedia article: http://en.wikipedia.org/wiki/Small_Busi ... n_Research" onclick="window.open(this.href);return false;

Most VR companies are very small entities, and most of the big jumps you have seen in HMD tech? Funded by SBIR money. In theory, SBIR grants allow them to have tons of money that they use to develop cutting edge technology, then puts them in a position where they can market it commercially. Problem is, it just creates an environment where you have a small company working for several years, burning through government money to develop new HMD tech, and once they finish development... Well, then what? You have no money, no way to attract investors, and a product that you only have the ability to produce in low quantity for high cost (Essentially a prototype). It is a massive jump from that to mass market, and most of the companies do not make the jump, so they sell to the military, where they can charge $50,000 per unit and nobody bats an eye. Realize that these things DO cost a lot to make. The parts alone can cost thousands to tens of thousands, and paying the employees adds thousands more.

Big companies like Sony COULD do it... But they do not see a market. And there is at least one person who wants to develop for the mass market: Me.

Side note: I might split this thread if it turns into a "Why good VR is not cheap" thread, it would let us get more off topic.

[Okta]

My question regarding the optics is, because the screens are so large and their centre of view is much wider than the eyes normal convergence point, how is the path of the eyes vision bent outward to compensate? You know what i mean, the reason i cant use 2 4inch plus screens without resorting to yucky prisms or mirrors.

[ERP]

@ERP: Do you mean how they are configured internally? It is actually a very simple setup. Two small screens side by side, and a pair of lenses in front of them. I would elaborate on the lenses, but their design is one of the things I specifically cannot talk much about.

Interesting how is the chromatic aberration?
The leep optics use a 3rd lens to correct it, doing it with 2 lens would certainly be nicer.

[PalmerTech]

@Okta: The control box handles that part. It puts the center of the view way over on the right side of the left panel, and the left side of the right panel, cutting a bit of the view off on each side. This is a lot like what our own eyes do, they do not have 100% binocular overlap. We can also adjust the position of the image in software.

@ERP: I worded that badly. There are not two lenses for each eye, there are two lenses total! A single lens for each eye, it really helps with the weight. As for chromatic abberation, there is essentially none. On the faaaaar edges of the display there is, but it takes a lot of effort to even look over that far!

[Okta]

@Okta: The control box handles that part. It puts the center of the view way over on the right side of the left panel, and the left side of the right panel, cutting a bit of the view off on each side. This is a lot like what our own eyes do, they do not have 100% binocular overlap. We can also adjust the position of the image in software.

Cool, anyone know of an effective way we can do this? Im eyeing off some vitrolight 4.8inch 1024x600 panels...

[PalmerTech]

Depends on what you want to drive it with. Your options are:

1) Use a custom engine, you can define the output parameters however you want
2) Use software like Nthusim or Warpalizer. Fine for videos, but I am not sure how well they work with current 3D drivers
3) Buy video receivers that support image shifting, $2000+ per channel
4) Program an FPGA to do all the scaling work (Just as expensive)

Check my other thread I just posted, though.

[cybereality]

Wow! That looks amazing. I'm jealous.

I have one question: Does it actually feel like you are INSIDE cyberspace? Everything I have tried appears like you are looking at a screen floating in space.

[Tone]

@Palmertech - sounds like you're having fun... and the perfect guy for the job.

@Aphradonis - Wide FOV HMDs are in niche markets because there doesn't appear to be a market big enough to scale the manufacturing costs down to consumer level. Over the years a number of firms have taken a run at the consumer market and ultimately crashed and burned. In particular, the lenses don't scale down in cost very well. A consumer HMD must have plastic, not glass, lenses for safety reasons and because quality glass is expensive. Wide FOV implies a large diameter lens which is difficult both to design and to manufacture in this short a focal length. The manufacturing is difficult because the lens is thick, and aspheric. Plastic lenses are injection molded and all sorts of problems occur as the plastic cools in the mold, primarily that the final shape is unpredictable. The thicker the lens, the less likely it is that you'll get consistent product. The usual solution is to toss the bad lenses in the trash, but the remaining good ones become more costly. To achieve low distortion and low chromatic aberration, the lens is a custom, coated, aspheric design; very tricky stuff. The molds are an art unto themselves, as they also pre-compensate for the shrinkage and expansion of the plastic as it cools. A mold for a lens of this quality and size can easily run six figures, and that's somewhat of a crapshoot, because it's quite possible the first (and second) attempt won't work. Specifically, I've been down this road with lenses of about the same diameter, but a focal length about 5 times greater (much thinner, easier lens), and our rejection rate was about 50%.

Inventors usually patent the clever things that go into such a HMD. Eric Howlett held patents on the Leep optics, Lenny Lipton had one relating to a key aspect of the big lenses I worked on. Royalties add to the cost; hundreds, maybe thousands of dollars per unit. Howlett's and Lipton's patents have expired, but for any HMD of quality, there's still a minefield of intellectual property out there.

So... if you're frustrated with the state of consumer VR technology, go raise a few million $, and make us all a great HMD. Step right up!

[PalmerTech]

@Cybereality: It absolutely does. Keep in mind that 150 FOV is more than you get wearing most combat helmets, and I am sure soldiers still feel pretty immersed in the battlefield. I still need to do a writeup on our motion capture stage and VR engine, sorry for putting that off! I will cover immersion in depth when I do that.

@Tone: Thanks! Definitely having fun, plan on sharing as much as I can. Thanks for the great explanation on lenses and scaling costs, said much better than I ever could have.

[cybereality]

Would stereoscopic augmented reality count as ortho-stereo? I mean, 3D objects can appear at actual life-size and in the proper perspective.

[PalmerTech]

Not thought about that before, actually! But yes, I think it could be.

[brantlew]

I would elaborate on the lenses, but their design is one of the things I specifically cannot talk much about.

Those glasses seem really impressive. Those links are helpful, but I'm still trying to get my head around LEEP systems a bit more. So in a typical system the first lens basically grabs all the light in a wide FOV but warps the perspective. The second lens then attempts to "undo" the perspective warping and direct the light into the eye at a more realistic angle. Is that "roughly" correct?

But that is a completely physical system. Can software pre-transform the image so that the second lens is unnecessary? Basically if you design a custom lens/warping-transform-function combo such that the light enters the eye correctly through just a single lens then that would be a much more compact system. The lens design and transform would be heavily researched and sensitive intellectual property. Am I on the right track?

[PalmerTech]

Actually, LEEP lenses are pretty simple. The goal was to get as much magnification as possible, distortion of all kinds was mostly ignored. You see, LEEP lenses were originally designed to be used with a LEEP camera, which captured distorted images itself! When you fed the distorted images to the distorted lenses, it all got cancelled out, resulting in a correct image. For VR purposes, people pretty much just ignored the distortion, since it is only strong on the very far edges of the lenses, out of direct view.

As for warping, there is already software out there like Nthusim or Warpalizer that can do all the work. The Wide5 lenses are unique in that they achieve the high magnification they need with a very short focal length, and a minimal of distortion.

[WiredEarp]

To me, the difference in immersion between narrow and wide FOV (in 3D stereo). Narrow is like looking into a pool of water, while wide is like being submerged in the pool of water. This is the best way I can describe high FOV VR.

I don't really consider 45 degree FOV to really even BE VR. Its just like watching a screen, you never (well, hardly ever) lose the sensation that you are watching a screen, rather than looking around inside an artificial reality like with wide FOV.

I'm eagerly awaiting the Sony HMD, as i'm hoping someone will come up with replacement optics to allow decent FOV finally. This Wide5 sounds awesome!

[cybereality]

You know, even a desktop monitor can be pretty immersive if you use a decent sized fresnel lens. Its all about the optics.

[WiredEarp]

True, but you never get the sense of being underwater and the HMD being the mask you are viewing the world through with a desktop monitor. Fresnel monitor setups are more like sticking your mask in the water and looking down (sitting on the surface still) rather than being totally enveloped in the 'water'.

I like my multi monitor flight sim setup, and my 3d monitor with head tracker, but neither are as good as a
proper high FOV HMD.

However, I was tossing up the idea of getting a 720P projector and sitting close to provide the high FOV. I think that should give sufficient FOV - but I dont think i'll bother now the Sony HMD is coming out.

Fresnel systems are actually pretty immersive - but I don't think you need them if you have stereo 3D? Doesn't stereo 3D provide pretty much the same effect, along with popout if required? Although, stereo 3D works best at close distance, and Fresnel lenses simulate long distance - but the effect seems pretty similar from what I can remember?

[cybereality]

Yes, but you can combine fresnel and 3D. Double the fun. But true, its not going to match a good HMD with headtracking and wide FOV.

[brantlew]

Actually, LEEP lenses are pretty simple. The goal was to get as much magnification as possible, distortion of all kinds was mostly ignored. You see, LEEP lenses were originally designed to be used with a LEEP camera, which captured distorted images itself! When you fed the distorted images to the distorted lenses, it all got cancelled out, resulting in a correct image. For VR purposes, people pretty much just ignored the distortion, since it is only strong on the very far edges of the lenses, out of direct view.

As for warping, there is already software out there like Nthusim or Warpalizer that can do all the work. The Wide5 lenses are unique in that they achieve the high magnification they need with a very short focal length, and a minimal of distortion.

OK, so the Wide5 lenses are custom cut to optically reduce distortion. Could you "in-principal" just use a generic wide angle lens but process the video so that it was pre-distorted in such a way that when it passed through the lens it exited undistorted? So you end up with a GPU + generic lens producing the same effect as a specially designed lens.

[PalmerTech]

Yes, you could. The problems with that are:

1) You still have chromatic distortion, which is next to impossible to correct for in software currently
2) You are probably going to have issues focusing on the edges. This is a minor problem, and not really that bad.
3) You still need really high magnification

The problem is, in the end, that you need very high magnification, at a very short focal length, in a very large lens. Pretty tough order! The LEEP optics basically do exactly what you describe, though: They just ignore distortion, and you deal with it in software. But LEEP optics need 3 lenses per eye to achieve that, which is a lot heavier than a single lens.

@WiredERP: Here is a copy paste from my other thread, regarding HMZ-T1 upgrades:

As far as the HMZ-T1 upgrades, we will see. I have access to some lenses that give 100 FOV out of a 0.97" Kopin microdisplay, but using them with the 0.7" Sony panels would not be nearly as impressive. It should be possible to push the FOV to 60 or 70 diagonal, but the Sony optics are probably much nicer, clarity wise. We will have to see, I have a few ideas. Either way, it will not be a removable lens, it would have to be an internal modification. An addon lense for the HMZ is theoretically possible, but it would be a lot of optics to cram in a very small space.

[nrp]

Yes, you could. The problems with that are:
1) You still have chromatic distortion, which is next to impossible to correct for in software currently

Chromatic aberration is actually pretty straightforward to counteract using a fragment shader, though I agree that it is preferable to design optics to minimize it when possible.

[PalmerTech]

Oh, I know you CAN do it. But only in your own programs, there is no way (That I know of, am I wrong?) to adjust existing software. Another problem is that at lower resolutions/high magnification like this, you do not have as much flexibility with moving pixels around as would be ideal.

[nrp]

Oh, I know you CAN do it. But only in your own programs, there is no way (That I know of, am I wrong?) to adjust existing software. Another problem is that at lower resolutions/high magnification like this, you do not have as much flexibility with moving pixels around as would be ideal.

Ah, yes, sorry, I assumed that high FoV hardware would be going along with custom software. Doing it to existing software would definitely be extremely difficult. Hacking up xrandr or something might do it on Linux, but I don't know if its even possible on Windows.

Edit: Or a passthrough box that does the transform, but that is difficult, expensive, and adds latency.

[ERP]

@ERP: I worded that badly. There are not two lenses for each eye, there are two lenses total! A single lens for each eye, it really helps with the weight. As for chromatic abberation, there is essentially none. On the faaaaar edges of the display there is, but it takes a lot of effort to even look over that far!

Interesting, do you know how big the screens are?
I assume they offset the images to correct for the angular mounting rather than include some sort of prismatic correction?

looking at pictures of the housing, and assuming the screens are the size of the front, I'd imagine you'd need at least a 60mm focal length lens, possibly a 40mm.
I did some experiments with a single 60mm on an eyephone screen and the distortion was a lot less than I had anticipated.

Certainly an interesting approach.

[PalmerTech]

Actually, the lenses are mounted angularly, too. That is one of the ways they achieve such a high FOV! But there is a lot of fancy image processing to make it work properly. Getting the Wide5 to work well with "normal" games is pretty hard!

I know quite a bit about the internal components of the Wide5, but I should really not talk about it, because some of the details may have been told in confidence.

[bobv5]

Have you tried the gear at work with normal games? How was it?

[ERP]

Actually, the lenses are mounted angularly, too. That is one of the ways they achieve such a high FOV! But there is a lot of fancy image processing to make it work properly. Getting the Wide5 to work well with "normal" games is pretty hard!

I know quite a bit about the internal components of the Wide5, but I should really not talk about it, because some of the details may have been told in confidence.

That bit I'd guessed. They'd pretty much have to be to be close enough to the eye to get 120 degrees.
I'm actually less interested in existing games, I think they'd be nice but not essential.

Can you say how big the screens are?

[Bishop51]

Actually, the lenses are mounted angularly, too. That is one of the ways they achieve such a high FOV! But there is a lot of fancy image processing to make it work properly. Getting the Wide5 to work well with "normal" games is pretty hard!

Granted, I have very close to zero idea of how Wide5 works but I wouldn't imagine it would be too hard to get commercial game titles to work on your PR3 with some competent pixel bending software http://orihalcon.jp/projdesigner/" onclick="window.open(this.href);return false; . Just wondering if that would help the cause on gaming in Wide5.

[ripcurl123]

Interesting to see. Now why is 2 small screens and 2 small fresnel lenses and some plastic so expensive?

i second that freke1 its ridicalous

[PalmerTech]

Interesting to see. Now why is 2 small screens and 2 small fresnel lenses and some plastic so expensive?

i second that freke1 its ridicalous

Read the rest of the thread, it is not ridiculous.

[ripcurl123]

Wow! That looks amazing. I'm jealous.

I have one question: Does it actually feel like you are INSIDE cyberspace? Everything I have tried appears like you are looking at a screen floating in space.

good question !!

[mwarren]

Price: You essentially cannot buy one. At one point, several years ago I believe, a member here got a price of $35,000 quoted, but even if you had that money in hand, you would not be able to buy one. At the moment (As far as I know), there are no spare units just lying around waiting to be sold, so it would have to be a custom order.

The US Army's Rock Island Arsenal spent $32,500 on a Fakespace Wide5 in 2007.

The Wide5 uses two 60Hz, 1600x1200 pixel displays combined with a 6-DOF Intersense IS-900 head-tracker and their optics provide a field of view 150°×88°. The optical axes appear to diverge by 25° to gain extra peripheral vision in the same manner as LEEP Systems' Cyberface2.

The SBIR also publishes details about more of Fakespace Labs' awards to give you an idea where Wide5's might be used.

[profvr]

I've used this HMD a few times in a couple of different labs.

One thing Palmer hasn't mentioned, but is really distinctive about this HMD compared to other wide FOV displays, is how bright the display is. You really feel as if outdoor scenes are outdoors.