Despite them doing their best to cover it up with quick editing and cherry picked angles, it's clear this has very low depth resolution and foggy picture quality.
If this only uses light, how would it deal with occlusion? It would be impossible to put a dark object in front of a lighter one, right?
I am on the Looking Glass team on Volume. I can answer you.
As far as resolution: each slice is about 104px high resolution in it's current state, the width is about 1500px in resolution.
The current depth resolution is 10. Because Volume currently has 10 slices. We do use some tricks to reduce the apparently low depth resolution, for example we allow a dev to set settings to 'overdraw' the slice so slices draw parts of their sister slices, which are then blended/interpolated with shaders. It helps a lot, and in fact our tests with more slices don't seem to do much other than increase the quality and viability of wider viewing angles. At near straight on viewing angles (within 15 degrees of the center), more slices don't seem to affect the quality of the display much (interestingly).
As an additive display, you are correct, a dim object in front of a light one wont show well... it really does look a lot like 'princess leia' in the 'obi-wan come help me' scene.
There are 2 ways we found to mitigate this.
1) We allow devs to set a 'black point' which adds the desired value to all geometry, thereby having a way to distinguish 'black' from 'void'.
2) Certain applications benefit a lot from not rendering slices at all but drawing everything as a normal 3D scene and then using a shader reading the depth buffer to determine the slice of a particular pixel - hence anything in front of something else causes a 'shadow' in subsequent slices (so it -CAN- occlude in this way, but by sacrificing the view from the back side). Things like scanned human heads look very good with this method... they honestly look like ghost heads suspended in space inside Volume.
Anyway, I hope it answers your questions.
(I am the main tools dev).
Thanks for the answers! I think it looks like interesting tech and it could be fun to play around with it, despite the limitations.
But I really think you should reconsider the messaging around the product. I'd imagine most of your potential customers would be tech-savvy enough to ask the questions raised in this thread. Right now your marketing material is making us suspicious; it looks like you have something to hide.
I think you'd be much better off if you were transparent about the technical limitations, so as not to be mistaken for another Cicret Bracelet (https://youtu.be/KbgvSi35n6o). Getting that out of the way would allow you to focus your story on the possibilities, why this tech is cool (because it is!).
Also, judging by your description some of the material on the site is very deceptive, such as the animation with the horse. It appears to be very high resolution, there's no "slicing" even at acute viewing angles, and it has dark objects occluding lighter ones. That's not really OK.
Thanks for the feedback. Knowing precisely what threw you off is super valuable to know, and I will directly tell the team members in charge of marketing about it.
I can also tell you for sure that there was no intention to deceive or misrepresent. In fact, anyone at all who is in the NYC area is welcome to see and judge Volume for themselves at our Brooklyn lab. We have a 'library' set up exactly for that purpose - to let the public come see and play with it! :)
However, I also definitely see what you are saying (and I will push for the horse in particular to be changed or updated). If there are any other things that threw you off, please do post in reply.
For anyone interested to drop by the lab and see Volume for themselves feel free to email: future@lookingglassfactory.com (the lab is at the North tip of Greenpoint Brooklyn, across the bridge from the Vernon-Jackson stop on the 7 train)
Yes, kinect 2 has around 200,000 pixels from the depth camera (512x424). I believe they said this thing only has 3,000,000 voxels. That's like 145x145x145. Barely enough to have a few voxelized Windows 3.1 desktop icons bouncing around.
It definitely seems like just something cool to have sitting around at this stage, like a lava lamp or something.
As long as they can keep improving the res, it could eventually be useful, but will they get there before unobtrusive AR glasses?
~1500px X 104px X 10px which probably is enough to be interesting and of course this is a gen 1 device. I suspect bandwidth is a major issue as this is already close to 1080p pixels.
They say it's not a parlor trick, but it's nowhere near true holography or anything like it. True holography records the object's wavefront as a flat interference pattern, encoding the phase information from the object (and also creating a very precise model of its surface), creating a diffraction grating which can then be used to replay the identical wave when illuminated by a suitable reference source of light.
This is the only known way to create true images with real depth of field and parallax, and requires ultra high resolution to both record and display the microscopic interference pattern.
There is a good book https://www.crcpress.com/Ultra-Realistic-Imaging-Advanced-Te... that has the mathematical basis for which the real principle of holography could be applied, given a capable enough display technology, to re-create an exact replica of the light field/wavefront that is identical to what we see in real life. Of course, a single static image with sufficient resolution of about 3"x5" in real life contains over 100 GB of data. The information carrying capacity of light is amazing.
None of the current VR/AR systems (including MagicLeap) use anything close to what is required, and we are still ways off in both display tech and GPU bandwidth to even generate a single, static image in a consumer product.
There's a small company in Austin, Zebra Imaging, which generates CG holograms on big pieces of photographic film.[1] Those are real holograms. The military buys them as 3D models in a flat, portable form, and they're sometimes used in place of architectural models. Some people still take photographic holograms using a big flash laser. (This was a thing about 40 years ago. One guy still does it.) Almost everything else called a "hologram" is about as fake as a "hoverboard".
Someday someone may make a display with light-wavelength-scale resolution and display real holograms in real time, but that hasn't been done yet.
I found this interesting. Too bad the prints are like $250/$500 each depending on whether they are monochrome or color. Wasn't easy to find out either, what with their expired webshop domains and all.
They look pretty good with the right lighting, but objects must never be clipped by the papers edge. So you have to keep your 3d inside an imaginary pyramid on top of the medium.
I saw a comment today on HN showing voxiebox[1] which seems to have a better volumetric display, unlike this one which has a very small number of discrete layers.
I think it's a spinning plane with a projector beneath. Their limitation is in the DLP projector's refresh rate. Seems they also may have extra limitations in colour space, since most DLP projectors use a rotating colour wheel.
It have to agree with the parent, it definitely looks like a display or a projection surface oscillating in the vertical direction.
I'm really curious if they opted for a display or a projector. A display is heavy but it might be possible if they use a vacuum. A projection surface on the other hand can be very light but you have to calibrate the optics and it _has_ to have a display and not a DLP chip, etc.. it's a really interesting problem to solve.
Looks a front-surface OLED display that's vibrating up and down on its Z-axis very fast.
So the number of layers would be bound by the frame rate of the display, the amount of vertical travel, and the speed of the motor driving it up and down.
I'll start by saying this is a really interesting device, and a very impressive technological feat. However, I have some major criticisms too:
The only advantage I see to this vs. something like HoloLens is that it doesn't require a head mounted display. Unfortunately for this project, that's not as big a barrier for serious companies as they need it to be. In time I think this "advantage" will also dwindle due to shrinking component size, etc.
The price is also considerable - it's about a thousand dollars cheaper than a HoloLens (I'm going to keep going back to HoloLens because to me they're directly comparable.) But the thousand saved also comes with some significant drawbacks:
- The notion of collaboration seems to be non-existent with this device - I don't see them showing how that might work based on their promo clips. Seems to be single user driven.
- It has a back-face. How do you render something like text so that it's readable from all perspectives?
- Control appears to be limited with no standard input. HoloLens tracks the users hands, by comparison.
- Does it have a top on it? At that point, what really separates it from a 3D TV? Seems like you'd only get limited perspective translation.
I'm having trouble coming up with really practical applications for such a device, that wouldn't be better suited by a head mounted personal device. I fully support the effort if they find their niche, but coming from a VR/MR point of view, I can't really see any real advantages. Please enlighten me! I don't mean to be a downer!
Full disclosure, I'm working with the company right now. I made the Unity Asset Tutorial Video as well as the rhythm game Rhythm Reach (and am currently taking a break from prepping for our launch party).
Also, take what I'm going to say w/ a grain of salt as I've been working in Volumetrics on and off for nearly 10 years with my personal project Lumarca: http://www.albert-hwang.com/lumarca/ -- so I've been drinking the Koolaid for a while now...
Volumetric Tech is a completely different medium, full of it's pros and cons. Many of the things people are bringing up in this thread are totally true and often infuriating when you start working in this space. You get x-ray vision (for better or for worse). It's impossible to meaningfully document (video always smashes it back into 2d). A focus on voxels make traditionally 2d content a weird problem (like text). And, to be sure, supportive tech for this isn't nearly as advanced as the stuff supporting VR. In terms of engineering, this creates a tough balancing act between price, visual fidelity, and scalability.
So -- why do it?
I do it because it's the only form of digital 3d media that exhibits passive physical presence. It's viscerally present in a way that no other digital media is. When people see a truly volumetric display for the first time (in person), 90% of the time they're totally floored. This is true even of my other volumetric display that had much much worse visual fidelity. The content is "worldlocked" in a very real way.
You also bring up the very real problem of practical applications. We're also exploring these questions (medical visualization w/ DICOM integration, songrams, games?). While this display is far from perfect, we're hoping that the low barrier to entry (a little $ + Unity) + some community presence, we'll make the environment friendlier for finding those solutions.
Anyhow, if you (or anybody on this thread) visits the NYC area, please drop us a note to come by to see it for yourself in person!
The description is rather vague. It claims 2 million voxels and no moving parts. That's resolution of maybe 200 x 200 x 50; the depth dimension seems smaller than the others. Is is some optical system which projects on 50 plates (maybe less), or something like that? It looks like they're somehow optically remapping a projector to multiple planes at lower resolution.
There have been depth displays before - vibrating mirror devices (a mirror mounted on a subwoofer driver) and such. Today we have enough GPU power to drive a volumetric display, at least at the modest resolution of this device.
But it's awfully low-res. Once you get tired of the 3D effect, it's going to be painful.
If you look at the rendering of the device in the video [1], I think it's pretty clear what they are doing. The device has a projector at the bottom that projects onto 10 angled surfaces that scatter the light in all directions. One row of pixels in the projector maps onto one row in one of the layers (or maybe they use the display columns, which makes more sense resolution wise). You can even see the angle of the surfaces in the image.
The resolution does seem super low, but I think they're trying to target the low budget market where it might make more sense. Their "How It Works" page says this:
> Volumetric displays have been around in research labs and a few small companies for decades. But they have always been extremely expensive — a staggering $100,000 per unit.
> Volume is nearly 100x less expensive than the systems that have come before.
Of course, their lowest price point is still $999, so I still wouldn't call that cheap.
It would be interesting to find out what the dimensions of this are in voxels. The 2 million voxels makes me think this is based on a 1920x1080 projector.
A graphic on https://www.lookingglassfactory.com/how/ showing the obligatory macbook has its display divided vertically into what looks like 10 slices - so perhaps 1920x108x10?
That would seem plausible as to what the photos on the site represent, but some of the 3d renders use a much higher vertical resolution for the display.
I believe the video is mostly concealing one important fact: the resolution in the z-axis is very low. That's the only way to explain the fairly high-res images in the demo when there are only 2 million voxels.
I'd guess the device is made of something like 20 stacked panels, each with a 300 x 300 resolution.
Fun idea, but the resolution is absurdly low, and as an emissive display it won't be able to block light (show dark things) without trickery or some drawbacks (like tinting the view glass, or requiring all lights to be off).
"people of all ages can come together to experience something that’s future AF."
Right, but it's the same reason I got a RealD television instead of a shutter frame one. You're not gonna give a roomful of people (especially if they're children) shutter glasses. Too expensive by half. RealD glasses are easy to come by and when the dog eats a pair (or a couple, in my case) it's an inconvenience.
Headsets are great when you're only entertaining yourself. Terrible when you have friends you're entertaining.
I think it's a very similar effect, except the 'reflection' surface is also the projection surface, across 10 panes. They've also obviously used some interesting optics for projection surfaces. Might be light guides.
It looks like multiple such displays stacked up, for true depth. Each of your eyes will see a slightly different image because it has real parallax, and each person around a display will see from a different angle too.
You're right. Each display uses diffuse instead of specular reflections. That way each pixel that gets projected onto the screen shines in all directions.
If this only uses light, how would it deal with occlusion? It would be impossible to put a dark object in front of a lighter one, right?