Support NeoGAF

[The Mad Draklor]

Never forget: https://www.youtube.com/watch?v=p2qlHoxPioM

I'm still sour it didn't half turn out like it was presented in the videos

The first time I saw that video, I knew the Kinect would not have most of those functionalities. I can't believe so many people actually believed it could do that.

 

[wwm0nkey]

Well it IS a Microsoft product, so you'd be a fool to expect anything else.

Hey man, the Zune was really good at least!

 

[Zaptruder]

I'm sad that people see kinect as snake oil, a worthless product that flat out doesn't work. It DOES work, just not the way they are using it.

It's like trying to eat peas with a knife, and then declaring the knife a worthless tool. It has utility, but it is misused. And now people will shit on it forever, which honestly jeopardizes it's utility in the fields where it is relevant.

It's called "poisoning the well." Microsoft poisoned the well on 3D scan cameras. They shouldn't be used as input, they should be used for what they are - a means of scanning an object as a 3D mesh.

They can and should be used as input. Full body motion capture has amazing potential for social VR.

It's just the way that they used it was atrocious. Kinect is a 3D input at a total mismatch from 2D display paradigms. This turned issues like navigating menus - one of the most basic tasks of computing interaction into a chore.

It's still not great even with a 3D input paradigm... but as a supplementary input system for proper motion controllers has great potential. Certainly far better than strapping motion sensors all over your body.

The latency issue can be worked on - isn't intractable. And when paired with good low latency primary tracking areas (head and hands) can be quite servicable even as it stands (i.e. 60-100ms for tracking torso and legs, reinforcing arm and hand locations is really not too bad).

 

[watership]

OP do you work for Gawker or something? You pick a thread title that says something extremely accusatory, when the source video has a totally different tone, then say watch the video. You can't have it both ways man. This is on the Verge too, which has slipped into clickbait since Josh Topolsky left, and MS for showing the huge wide angle vista, when the demo units aren't there yet.

This is the full article.

http://www.theverge.com/2015/5/1/8527645/microsoft-hololens-build-2015-augmented-reality-headset

People often imagine virtual and augmented reality fusing, but with HoloLens around, the two start seeming very distinct indeed. Its images are astonishingly good, on a level that VR's magnified screens will probably never match. It's smaller than any virtual reality device on the market, partly because it usually doesn't need to power an entire photo-realistic environment. Microsoft has put much more work into building things that people can use, not just things they can see. But it's hard to imagine how Microsoft (or anyone) could get the HoloLens projection system to support a field of view big enough that it can stop being distracting, let alone become immersive on a VR-like scale. As cool as HoloLens can be, it's firmly a product of today, not the future — at least not yet.

Also:

Typical Microsoft everyone :/

See? This is the shitposts you get with titles like that.

 

[Jomjom]

Hasn't this FOV issue been solved already? I remember seeing that one demo of the the sackboys for Morpheus that basically did AR that covered your entire field of vision. Or was that also just a small rectangle too? I'm not sure.

 

[hodgy100]

Typical Microsoft everyone :/

 

[Anton Sugar]

First off: this is nothing like Milo. Milo was just straight up fabricated.

This is overselling. Which isn't great, but it's a dramatically different proposition. The device does everything they're saying it does, just in an overly narrow field of view.

This.

 

[dose]

snip

But they also failed to show what happens when something appears in front of the virtual windows/objects. ie., it totally breaks the illusion. It was a very staged demo with careful camera angles so as to not show this flaw.

 

[Nzyme32]

It's amusing to note when GAF decides they can trust The Verge or not.

It's important to note a distinction here: the field of view/level of immersion from the demo is the "lie". The functionality is all there, according to impressions

The problem is the same as in VR and AR - the functionality of the headset is meant to bring a sense of believability which in turn aids the functionality and usefulness of applications / rendered tools / objects etc. If in either VR and AR you can not believe what you are seeing / it doesn't behave as you are expecting it to, the functionality also becomes flawed. Eg - If I create a 7ft screen showing a movie on a wall, the functionality is redundant when I can't see it until I am at a particular distance away, or I have to move my head constantly to understand what I am seeing.

I'd guess this is more so in VR where you are essentially trying to create an entire world rather than AR where you are creating specific objects and tools.

 

[Stimpack]

I'm more interested in what she said Hololens was, rather than what it wasn't. I honestly didn't expect much to come out of it, but to hear that it basically works and that FoV is the real issue is surprising to me. Time will tell if it holds up when it's out in the wild, but at least now it has my interest.

 

[Xater]

Not surprised at all.

Also MS just isn#t an innovator anymore. The last time they had some real impact probably was when Windows 95 came out.

 

[irrotational]

They demo they showed where they placed the floating screen on the wall also hid another drawback - the inability for hololens to create occluded pixels. Everything it creates is transparent. They placed the video window on a wall, because floating in space, it would look like stained glass.

I don't think that's true - hands-on reviews say the items are solid and occlude whatever is behind them.

 

[Krejlooc]

The latency issue can be worked on - isn't intractable. And when paired with good low latency primary tracking areas (head and hands) can be quite servicable even as it stands (i.e. 60-100ms for tracking torso and legs, reinforcing arm and hand locations is really not too bad).

Yes it is. 60-100 ms is inexcusable for our proprioception. Kinect skeletal tracking is inherently latent technology. It will never be overcome. It is not the solution for user input.

You have extremely high proprioception in your head.

 

[Guevara]

It's amusing to note when GAF decides they can trust The Verge or not.

Just to be clear: the entity in this story I don't trust is Microsoft.

 

[Krejlooc]

I don't think that's true - hands-on reviews say the items are solid and occlude whatever is behind them.

I have spoken with people who have tried hololens directly. The pixels don't occlude.

 

[SPDIF]

Microsoft isn't known for innovating, they're known for imitating. They'll either improve upon something that already exists (in part or in whole), or buy the person/company that actually innovated it. This shouldn't shock anyone.

Not surprised at all.

Also MS just isn#t an innovator anymore. The last time they had some real impact probably was when Windows 95 came out.

I don't understand these two comments, especially in this thread. You don't find HoloLens the least bit innovative?

 

[nib95]

Not surprised at all.

Also MS just isn#t an innovator anymore. The last time they had some real impact probably was when Windows 95 came out.

I don't think that's fair. Kinect, and even this Hololens are imo innovative. I think it's just the implementation that is somewhat less polished and successful than expected, or initially advertised.

 

[Overside]

Wat?

No guys, you dont understand, wait until you connect it to the power of the cloud!!

 

[InaudibleWhispa]

Hasn't this FOV issue been solved already? I remember seeing that one demo of the the sackboys for Morpheus that basically did AR that covered your entire field of vision. Or was that also just a small rectangle too? I'm not sure.

Morpheus has a screen inside, so to see the outside world like AR it must use a camera feed. Cameras capture a 2D rectangular plain, so yes it would be a smaller rectangle as far as I'm aware. You can view wide FOV fully 360 video feeds in VR that have been captured with specialist equipment, but not a real-time AR feed (at the consumer level).

 

[Bluenoser]

OP do you work for Gawker or something? You pick a thread title that says something extremely accusatory, when the source video has a totally different tone It's clickbait bullsshit.

This is the full article.

http://www.theverge.com/2015/5/1/8527645/microsoft-hololens-build-2015-augmented-reality-headset

But in the video it linked to, she clearly states that it's basically a lie because of how they initially showed the demo versus how it really works. Just because the written article doesn't say that exact same thing doesn't mean clickbait or accusatory.

 

[paulogy]

But they also failed to show what happens when something appears in front of the virtual windows/objects. ie., it totally breaks the illusion. It was a very staged demo with careful camera angles so as to not show this flaw.

Yes, and this would occur much more frequently if the demos were shown from the subject's first person perspective. I think this is why they have been showing these from a third party view, and the choreography to avoid breaking the illusion really is quite impressive.

 

[Anton Sugar]

We already knew about the limited FOV. However, there are still tons of impressions that use the word "mindblowing".

3 devs who got to use it at Build discussed and were very positive about it, even with mention of the limited FOV: "I am not a game developer and have actively shied away from developing games, but I left [build] wanting to design a game for this."

But they also failed to show what happens when something appears in front of the virtual windows/objects. ie., it totally breaks the illusion. It was a very staged demo with careful camera angles so as to not show this flaw.

This is a good question, but they have SOME way of handling this--I just have no idea how. From the very first demo, back in January:

At one point, there was a PC monitor right in the middle of our field of vision — this was to show how scientists do this today, and it's rudimentary compared with being able to move through the space in 3D. But it was also another example of how the real world and the virtual world commingled.

I held a call with a NASA engineer, and he talked me through the terrain. I squatted to look more closely at rocks, took snapshots of various rock formations, and even planted flags for points of interest. My jaw dropped when I ventured over to a PC in the room and started to experiment with the mouse. I pulled the mouse pointer off the screen and suddenly it was on the floor next to me, allowing me to set markers in the virtual environment. It’s everything I’ve seen in demonstrations from Microsoft Research before, but here it was on my head and working.

 

[AllIsOneIsNone]

Who the fuck at Microsoft decided that the "bring us up to let us down" concept was the way to go when marketing tech? If I was in charge I'd to the opposite. Present the thing as shitty but juuust cool enough to keep people interested, then the actual product will blow them away, haha.

 

[Anton Sugar]

Also MS just isn#t an innovator anymore. The last time they had some real impact probably was when Windows 95 came out.

lolwut

By all accounts, MS is going through a resurgence of new ideas right now.

 

[Jomjom]

Morpheus has a screen inside, so to see the outside world like AR it must use a camera feed and cameras capture a 2D rectangular plain, so yes it would be a smaller rectangle as far as I'm aware. You can view wide FOV fully 360 video feed in VR that has been captured with specialist equipment, but not a real-time AR feed (at the consumer level).

Ah I see. That's disappointing. How come nobody has mentioned that about the VR solutions? Seems like something to criticize equally as they are doing with Hololens.

 

[DrKelpo]

I mean, it's still 1 or 2 years before release, so of course they will improve some aspects of the device, but nonetheless... The straight up bullshit they pulled with kinect will not be forgotten.

 

[Three]

First off: this is nothing like Milo. Milo was just straight up fabricated.

This is overselling. Which isn't great, but it's a dramatically different proposition. The device does everything they're saying it does, just in an overly narrow field of view.

Kinect can theoretically do the Milo demo or the other fake scanning your skateboard demos. The problem is that its accuracy/practicality is not as advertised. Milo was overselling just as much as these staged/fake hololens demos are. They both just oversell to an impractical degree. I mentioned this exact issue in a past post before this article went up

http://m.neogaf.com/showpost.php?p=162130930&postcount=40

The practical use of a virtual screen that large goes right out the window with the limited field of view. The guy doing the demo would basically be seeing a small square of that screen. In practice that demo is bull designed to fool people because you would get a "bigger screen" watching a movie on your mobile.

 

[Cave Johnson]

I've always find AR fascinating in ways that even the VR doesn't. One, being occupied in the same space as reality, which brings me to this point:

Since AR is essentially interacting with digital interfaces in actual spaces, it is entirely possible to create new "controller" schemes out of everyday objects since they are heavily reliant on the AR input as opposed the mechanical aspects of the controller. Add with NFC or wireless capabilities to activate built in motors or such, I believe this would possibly make a controller tailored to any situation.

For instance, I can just get a normal white board that can instantly becomes a Wii-U like tablet interface simply the camera shooting at the interface (this means wider FOV cameras's outside our periphery to cover said area). In short, it virtualizes the control scheme and tethers it to whomever and whatever the player sees fit.

 

[Krejlooc]

Morpheus has a screen inside, so to see the outside world like AR it must use a camera feed. Cameras capture a 2D rectangular plain, so yes it would be a smaller rectangle as far as I'm aware. You can view wide FOV fully 360 video feeds in VR that have been captured with specialist equipment, but not a real-time AR feed (at the consumer level).

You can do so with consumer equipment like Leap Motion, albeit at a low resolution. Its currently in black and white because it is a pure depth scanning camera, no RGB offset, but dragonfly, its successor, will be able to do it in color with a higher resolution.

 

[dose]

This is a good question, but they have SOME way of handling this--I just have no idea how. From the very first demo, back in January:

I don't think your examples relate to that and tell us that they have a solution.
As I posted in the other Hololens thread, there's a point in the demonstration where the robot is BEHIND the guy, yet when he moves his hand close to the robot...

His hand should appear in front, but obviously doesn't because that would require Hololens to occlude in realtime part of the robot as his hand passes it.

 

[InaudibleWhispa]

Ah I see. That's disappointing. How come nobody has mentioned that about the VR solutions? Seems like something to criticize equally as they are doing with Hololens.

The strength of VR right now isn't in displaying real time footage of the real world like AR so the criticism isn't necessary. When displaying fully rendered 3D worlds or captured film footage with the right 360 cameras, the FOV is much greater. Still not perfect, but good enough right now.

You can do so with consumer equipment like Leap Motion, albeit at a low resolution. Its currently in black and white because it is a pure depth scanning camera, no RGB offset, but dragonfly, its successor, will be able to do it in color with a higher resolution.

Interesting. But still, that isn't something that first-gen consumers devices will be using (apart from enthusiasts tinkering with third party stuff), particularly Morpheus -- any real time AR sackboy demo with that would have to be a smaller rectangle camera feed, right?

 

[Mr. Giggles]

Hmmmm... Part of me wants to say "Told you so"

but...

If we consider the ground work MS has done we can expect great things down the line..

 

[Biker19]

See? This is the shitposts you get with titles like that.

Everyone else besides hodgy100 has been calling it as BS & have been calling out MS for it.

 

[Sjefen]

MS really knows how to sell it

 

[Anton Sugar]

I don't think your examples relate to that and tell us that they have a solution.
As I posted in the other Hololens thread, there's a point in the demonstration where the robot is BEHIND the guy, yet when he moves his hand close to the robot...

His hand should appear in front, but obviously doesn't because that would require Hololens to occlude in realtime part of the robot as his hand passes it.

Well, you can search for them in you want, but a number of those impressions said that everything in the room (bulletin boards, desks, things hanging on the wall, etc.) was covered by the Mars surface hologram except the computer terminal".

It could be smoke and mirrors, but I don't know how they did it.

 

[watership]

Everyone else besides hodgy100 has been calling it as BS & have been calling out MS for it.

The field of view is limited, otherwise it is stunning. Says every report you'll read on the internet. For a product that's a year to two out, "calling MS out on it" is nuts.

 

[Three]

Well, you can search for them in you want, but a number of those impressions said that everything in the room (bulletin boards, desks, things hanging on the wall, etc.) was covered by the Mars surface hologram except the computer terminal".

It could be smoke and mirrors, but I don't know how they did it.

Well they could really easily do it on a static object that relates to a position in virtual space especially with one capable of relaying some position information, IR markers or Light Gun type tech. The problem comes from moving objects without defined positions. with a good realtime 3D scan of the area it is possible.

 

[Zaptruder]

Yes it is. 60-100 ms is inexcusable for our proprioception. Kinect skeletal tracking is inherently latent technology. It will never be overcome. It is not the solution for user input.

You have extremely high proprioception in your head.

Good thing you're not looking at your legs and torso most of the time. It's also why I say it should be supplementary tech. The primary trackers should still absolutely be the motion units in the HMD and in the controllers.

Full body motion capture is useful for avatar gesture, as well as simply full body motion capture (animating model data with your own body/acting).

Skeletal tracking has higher latency because of the processing overhead. It's certainly conceivable to get that processing overhead down low enough (20-50ms range) with continued cost improvements in computing.

So... the current 60-100ms isn't ideal, but it can still be useful. It's also very far from the 300ms of latency you'd need for your brain to disown the motion.

 

[KemoSabe]

Anyone know if there is a Video about this?

 

[dLMN8R]

So basically all the impressions are saying that the room mapping, latency, and all of the actual impressive parts work extremely well.

But the field of view is the problem.



That sounds.....pretty awesome? That what everyone expected to be the most difficult parts of HoloLens are working astoundingly well, and the biggest problem seems like one that could be easily solved with a revision or two?


(disclaimer I guess, I work at Microsoft, but not on HoloLens. I'm just going based off of the impressions I've read online including The Verge)

 

[ironcreed]

I am just not into any of this gimmicky stuff, including VR. Just me give me some great games to kick back with.

 

[cartoon_soldier]

http://techcrunch.com/2015/04/30/hololens-is-real/

While I originally assumed HoloLens — like most other augmented reality applications — needs some kind of QR code or markers to know where to place objects, it can actually recognize the world around you and then allows you to place objects where you want them. Using the Unity physics engine, it can interact with them.

For the grand finale, we added a new object to the scene that would explode when one of the balls hit it — and which then revealed a cavernous world underneath the floor with a stream and animated birds. That was obviously the most impressive demo and really showed the potential of HoloLens. You could even drop the balls into that cavern and see them bounce 20 feet beneath your feet.

At the end of the session, I came away very impressed. When I first heard about HoloLens, I thought this was a technology that was still very far away from being production-ready and I assumed that the demos Microsoft showed earlier this year were simply well-staged and had managed to pull the wool over the assembled tech press. Now, I wouldn’t be surprised if Microsoft started selling HoloLens within a year (the company, of course, won’t say when HoloLens will go on sale or at what price).

So yea, tons of positive impressions from people who have tried it out first hand too.

yes, FoV is limited. That doesn't mean the technology and what is there is bad.

 

[Krejlooc]

Good thing you're not looking at your legs and torso most of the time. It's also why I say it should be supplementary tech. The primary trackers should still absolutely be the motion units in the HMD and in the controllers.

Full body motion capture is useful for avatar gesture, as well as simply full body motion capture (animating model data with your own body/acting).

It doesn't matter if you are looking at your legs or torso most of the time, however. Proprioception is what happens when you aren't looking at that body part.

You don't need skeletal tracking for avatar gesture, merely 3D scan. And, as a mocap utility, Kinect is terrible. It simply isn't the optimal solution. Something like forward kinematics like PrioVR is a much better solution.

 

[Three]

The field of view is limited, otherwise it is stunning. Says every report you'll read on the internet. For a product that's a year to two out, "calling MS out on it" is nuts.

There is nothing nuts about calling out MS regarding fake staged demos even if the product is a year or two out.

 

[Anton Sugar]

Well they could really easily do it on a static object that relates to a position in virtual space especially with one capable of relaying some position information IR markers or Light Gun type tech. The problem comes from moving objects without defined positions. with a good realtime 3D scan of the area it is possible.

True, good points.

 

[hodgy100]

Everyone else besides hodgy100 has been calling it as BS & have been calling out MS for it.

sorry guys. I'm just a bit salty that I called this out as too good to be true when it was initially revealed and got piled on.

I have a real love/hate relationship with MS

the stuff they do with windows and curating developers is for the most part excellent.

but their consumer electronics always has this bigger than reality air about it. they consistently get people hyped up only to disappoint them later on :/

 

[cartoon_soldier]

There is nothing nuts about calling out MS regarding fake staged demos even if the product is a year or two out.

Cherry picking one set of impressions vs. others is nuts.

 

[John Bender]

Shocking news!

 

[blakep267]

There is nothing nuts about calling out MS regarding fake staged demos even if the product is a year or two out.

How is it fake. The camera just has a larger fov than the hololense headset. The guy is seeing that stuff just in a smaller window

 

[dLMN8R]

There is nothing nuts about calling out MS regarding fake staged demos even if the product is a year or two out.

But what was staged about the demos?

The only thing "fake" about the demos appears to be the camera making it look like the person was surrounded by holograms while the actual user has a limited field of view of those holograms....which seems entirely reasonable?

The holograms would still be there if the user looked at them.