Looking Glass starts shipping its 8K holographic display
techcrunch.com
techcrunch.com
This shitty gif from techcrunch is infinitely more impactful in every conceivable way despite being a shit quality gif. https://techcrunch.com/wp-content/uploads/2018/08/Aug-22-201...
Looking Glass people, if you're reading this, you need to zoom way the fuck out, turn the lights on, and quit with the artsy bullshit fading to black every few seconds because it looks like you're hiding something. You cannot show pictures. You need to show the experience.
Too hard to connect the cables for that shot? Didn’t bring a computer to the photo studio?
Another option: short animated loop to create depth from parallax. But the key thing is it needs to show the monitor as a whole along with the image on screen, all moving together. https://imgur.com/eh5u6Gu
Seems similar to how pixel art worked, and tricks artists used to get a better image with CRTs. https://66.media.tumblr.com/8d2cf7adae94fde97d1a8c9cf78a46a2...
Watch this video, NOT their own demo if you want to be impressed.
They did not allow press to take images of the device when the display was on.
They knew images couldn't capture the 3D so they did what this company is doing and used a bunch of vague renderings to try to express how the thing worked.
Since it's probably the same underlying parallax tech powering both, I'd guess the reasonings are the same.
The later "New 3DS" introduced a face tracker that made the effect work continuously across most of the LCD's viewing angle for a single viewer.
Curious why they didnt go for a similar setup here? Though the looking glass display allows for multiple viewers.
It's not. Looking Glass is an advanced version of lenticular optics, which is 100+ year old technology, except used to create a multiscopic display(many views).
Nintendo 3DS is newer technology, known as parallax barrier. However, 3DS is only two views, which is pretty much only useful for a single viewer.
Even in that TechCrunch gif, I believe I can already see pixel borders on the specular shading of the top part of the h.
For an even clearer example, see the frog in full-screen at 3:31 here: https://youtu.be/-EA2FQXs4dw?t=211
The much larger “8k” version introduced half way through https://www.youtube.com/watch?v=-EA2FQXs4dw has some good tech detail, with the raw pixel mapping shown at 6:30. Calcs seem about right since raw pixel count was stated as 43 million, although it was implied elsewhere they split RGB sub pixels too (which sounds wrong I admit), so maybe 43 million divided by 15? Our eyes are more sensitive to luminance than colour, so maybe they did something there (however, not that I could see from the raw pixel mapping at 6:30).
100% / sqrt(2)
They need to throw up a _static_ 3D image that the viewer can easily understand and then move the camera. That's it. That's all they had to do, because literally the one thing that makes this screen special is showing different viewing angles, and they failed wildly.
The main thing that everyone wants is a lot of depth, a wow effect. Lenticular technology can only deliver that while your head is static. As soon as you move, you either see stripes move across the image or you need to introduce an unnatural amount of bokeh blur.
The reason for the stripes is that from your eye / camera, different parts of the image have slightly different angles, so slightly different subpixels are visible.
So if they follow your suggestion, they need to ensure that the final video is low quality enough so that you don't see the striping artifacts.
Here's a lenticular print that I had made at 4800dpi, so at a much higher resolution than what a display can hope to achieve
https://www.dropbox.com/s/10nquohksew1fhe/fertig.mts?dl=0
Note the strong blur to hide artifacts, yet you can still see some striping on the background and in the top right corner.
Here's a simulation of the best possible result that one could hope to achieve with 70lpi sheets and 1200dpi effective resolution, which should be close to what this display uses:
https://www.dropbox.com/s/1s69k5gmt1n4i5l/simulation_L.jpg?d...
For anything better than that, they need much more subpixels. 2x the pixels for half the banding width.
That being said, it also had limitations. 3D effects that extended beyond the display (like a tunnel, or some larger effect) lost that depth to me, and there's no up-down 3D, only side-to-side. The one I saw was also not as high-resolution as you might think, the effective resolution was the resolution/elements where each element was a particular angle the display was to be viewed from and the computer had to render the display from each angle simultaneously. The prototype I saw had the effect of looking a bit like 3d objects underwater.
Still, it was the closest thing I've seen to a volumetric display outside of a lab, and in the cases where it really worked (things inside the volume of the display) it was kind of jaw dropping.
edit here's the pouet page for the demo I saw https://www.pouet.net/prod.php?which=78756
This video is from a handheld camera that moves around the display that's helpful to understand how it works https://www.youtube.com/watch?v=US7hzM0a21E&feature=youtu.be
It reminded me a lot of those "3d" pictures you can get that have a bunch of vertical sections at slightly different angles. They go back to maybe the 70s. There's a funny name for them that I can't quite remember.
My understanding is that the "box" that sits on top of the display is purely cosmetic and helps sell the effect, and that the backboard of the display provides everything.
It’s lenticular printing - https://en.m.wikipedia.org/wiki/Lenticular_printing
But reportedly they execute that principle super well, gimmick part included, to the point it looks almost VR.
(I am sure they have thought of this by now ...)
One challenge here is the tradeoff between spatial and angular resolution. If you're generating a 4x4 lightfield starting with a 2160x1440 display, you only end up with 640x360 superpixels.
The company I work for (https://www.leiainc.com/) makes a display that can switch between light field and 2D, so you can have the best of both worlds.
45 angles for each pixel:
https://www.youtube.com/watch?v=-EA2FQXs4dw&t=4m18s
That's far more complicated than a lenticular sheet glued to LCD.
https://www.engineering.com/ARVR/ArticleID/17613/In-Through-...
Pure speculation.
https://hackaday.com/2018/11/21/supercon-alex-hornsteins-adv...
"since he happens to be head honcho at a holographic display company he can show us the result. Looking Glass Factory’s display panel uses a lenticular lens to combine the multiple images into a hologram, and is probably one of the most inexpensive ways to practically display this type of image."
At 6 min 30 in https://www.youtube.com/watch?v=-EA2FQXs4dw they show the pixel sub mapping - and it isn’t just vertical bars.
Edit: If they used a vertical grated sheet, there should be vertical lines at high contrast edges at a 45 pixel pitch. look at the green book at 6:22 and close-up at 6:27, and there seems to be vertical smear (not just what you might expect from horizontal smear). Example of a 3D lenticular print without the lenses to compare against: https://cdn.instructables.com/FKG/KDGQ/I0290WOV/FKGKDGQI0290... from https://www.instructables.com/id/Computational-3D-Lenticular...
One of the professors at CMU (who coincidentally also is writing the FM Towns emulator and has written a number of demos for Demosplash, of the first ever for those platforms), is a really cool guy who brings a library of almost impossible to play Japanese only PCs to the event. He's done some truly amazing stuff with some of them and they're a blast to play games on.
The folks that run it are great, including having a mainframe emulator running on a rPi as the sign-in system, and it should easily have 4-10 times the attendance. I really recommend it if you're just a few hours from CMU and can take the weekend. Even better, enter a competition!
IIR, the competitor who brought the Looking Glass display was a VR developer for Facebook, so there's some serious muscle that shows up.
Yea it was a lot of fun playing with the old machines at last year's Demosplash.
>One of the professors at CMU (who coincidentally also is writing the FM Towns emulator and has written a number of demos for Demosplash, of the first ever for those platforms), is a really cool guy who brings a library of almost impossible to play Japanese only PCs to the event. He's done some truly amazing stuff with some of them and they're a blast to play games on.
Soji Yamaka is a wizard at reverse-engineering these old systems. I am in awe at his drive and ambition in figuring out solutions to preserving these old systems. His submission for the 2019 Demosplash was one of my favorites. We had some great late night chats.
>The folks that run it are great, including having a mainframe emulator running on a rPi as the sign-in system
HA! I remember that. Geeked out with the founders for like an hour just talking about how they did it. Seeing that at the check in desk was my first introduction to how CMU students leave no stone unturned in nerdy awesomeness.
When I was in school, me and my fellow retro geeks would talk about writing stuff like this but we never actually spent the time to learn how to write software for old machines. We would just be trying to survive our own classes and then waste time on other things.
The CMU computer club takes the craft to a whole new level. I guess it is to be expected given that they are a top tier school. While the mainframe system they emulated was before my time, I loved the glow of the DEC monitor they used. Took me back to when I first saw one as a child.
>IIR, the competitor who brought the Looking Glass display was a VR developer for Facebook, so there's some serious muscle that shows up.
Ah man missed it! I don't recall any outside companies attending in 2019.
Do you attend every year or was it just a one off thing?
Does anyone use their 3D glasses that used to ship with "3D TVs" circa 2013? My Samsung TV came with a pair of active 3D glasses that were collecting dust. On the other hand we've had adoption of touch screens on handheld devices, it almost swooped the entire mobile market between 2007-2012 after iPhone's introduction. But the same thing didn't happen with Keyboard + Mouse input on a desktop computer. Infact, the market just exploded with new mechanical keyboard aficionados sometime around 2010, I still remember hanging out on geekhack a decade ago and now mechanical keyboards are everywhere.
We've seen in the past and we will see this in the future - a whole lotta focus on aesthetics, UI and presentation - i.e., cool graphics in video games, touch screens, holographic displays like the one in the article, AR/VR tech (magic leap?) without proper attention to content will lead to nowhere. Probably just make headlines.
Another thing is that people don't take ergonomics into account. Pretty much any sci-fi movie has people moving their arms about to interact with content. That would never take off in real life. Unfortunately, people like Elon Musk are hell bent on horrible UI/UX depicted in their favorite sci-fi movies and shoehorning it into Space craft (dragon capsule has all touch screen interface with literally no buttons, checkout the Everyday Astronaut channel's tour). Elon's vision about UI/UX is misguided through movies, it is embarrassing. The giant touch screen panel in a Tesla is the main reason I would never get one. I think he might put this 8k Holographic display as an option, please don't tweet him about it.
Sci-fi movies are the cancer of design. It is the victorian design equivalent of modern times, pure decoration. You can find traces of it in professional equipment, this thing looks like it was pulled from a space ship: https://images-na.ssl-images-amazon.com/images/I/61cGhQ0begL...
I wish smaller 3D computer monitors had caught on. At least VR headsets are still going strong and are even more immersive.
I have no experience with their 3d, only their HD (they call it HDX because marketing) which is perfectly fine for a streaming service.
The important ones, PS3 and PS4, do however.
VR googles have their own issues, because the lenses never seem to be perfectly in focus, and the resolution isn't high enough directly into the glasses.
In both cases you can choose to ignore the problems for a while, but at least in my case the eye strain builds up enough I doubt I could deal with it for even 4 hours a day on a regular basis.
I've never seen this on a monitor/display either. But I have an Epson projector from 2013/2014 that uses shutter glasses and does block 100% of the opposite eyes image. Because it's not a screen, it doesn't have to blank the image between frames, it just completely stops sending light from that frame.
A bit puzzling that you criticize all from ancient knob proliferation to a single large flat screen. Only alternative is small screens with deep menus, it’s own he11.
Nothing, as long as you have the space to put them somewhere sensible.
Google Glass sort of failed in the consumer sense but eventually has some adoption in the Enterprise space.
I have serious doubts about AR use in things like assembly instructions. Turns out, it is probably ok for training but if you do a task 30 times, your brain develops a memory for how to do it and you don't need to wear AR glasses. For maintenance techs, which is how Google Glass is marketed, it is too much of a hassle to put on the glasses, have a software team write the application and maintain it and then after spending $200k on this boondoggle AR project, what is the ROI? I really don't see it as of any value... maybe there is a positive ROI for airplane maintenance.
I am really not convinced. It looks like a solution in search of a problem. Are there any AR goggles used in manufacturing industry on a mass scale?
As I said, there are edge cases where AR/VR makes a lot of sense. HUDs in fighter jets for example are tremendously useful. But my complain is mostly about people wanting to jam some new tech in, take on a lot of tech debt, spend $$$ in a fortune 500 company only to find that there is no real problem to be solved.
Should we spend $120k consulting with a software company to develop an iPad based checklist? Not to mention yearly maintenance of that codebase. Printed paper + pen works just fine in 90% of the cases for 99.99% less cost! You have to ask - how many people are going through this "old school" paper trail? 100? 1000? 10 million? I've seen AR in manufacturing roadmap slides where there are like 13 workers and they are all with 25+ years of experience and the management wants to do their AR boondoggle. Frustrating.
For me, the interesting part about owing a Tesla is the realization of just how badly designed traditional cars are. Dedicated tactile controls are better in an automotive context for fiddling while driving, and the Model 3 implements almost everything necessary while driving as physical controls.
The two things that I fiddle with while driving that don't have physical controls are the stereo and windshield wipers. The stereo has a physical volume and tracking controls, I don't think a physical interface for any other part of the stereo would be much better than the touch screen. Windshield wipers should have a physical interface, in the Model 3's auto mode is great 99% of the time but when it isn't it's too hard to manually adjust and when you need to adjust wipers is precisely the wrong time to be fiddling.
It is fricking glorious. IMO, hardware controls > software controls. Big time and $$$.
All monitors today have no dedicated buttons for switching input sources. Instead we have a d-pad nipple that takes 20 seconds to pop up a menu, then dig into the menu shortcuts (thank god!) and then having to select input source.
Humans have regressed hard in last 30 years in terms of UX/UI.
That said, the OSD interface has nothing on the decade-old Dell that sits next to it, which even has the option to flip the interface for ease of use in portrait mode which is how I have it set up.
Physical buttons or dials for everything except for infotainment settings, but minimal with no clutter. Steering wheel input for cruise control/audio/wipers and the heads up display, while the climate controls are dead simple and easy to work without looking down. Infotainment has both touch and a context sensitive dial. Everything is grouped contextually and I never have to look away from the road.
I was especially impressed that a single dial (left, next to the lights) dims _every_ light in the entire cabin to the same brightness, across many different subsystems. Great design shields the driver from the complexity of the system.
I leave it on Defog most of the time (straight up) and change from there without needing to look. If I'm unsure a quick glance tells me the angle and I can make the adjustment without having my eyes on it.
Why do you believe this to be the case? I've been having a blast with the (fully wireless) Oculus Quest, especially games like Racket NX which involve constant swinging of arms (you're playing tennis, basically).
The main, enduring dimension VR adds to gaming IMHO is physical exertion (and this aspect is only really engaging with a standing experience... and then only enjoyable if wireless). I feel much better physically after the light exercise of half an hour of Quest gaming than I do after half an hour playing some PC first person shooter.
Glasses based 3D sucks. People who don't wear glasses don't want to wear glasses and find them unconfortable. People who do wear glasses don't want to wear two pairs of glasses and find it uncomfortable. Active shutter glasses give some people headaches from the flickering. If I could get perscription lenses with polarization for my TV's passive 3d, I might play with it... But I watched like one 3d blu-ray with the glasses and that's good enough for me. This display looks interesting because the viewer doesn't have to wear anything, but we'll see.
Touchscreens on mobile works because it's cheaper to build than a number pad, and way cheaper than a keyboard, it's cheaper to extend the touch screen so they don't use any real buttons on the front of most androids. The flexibility is helpful for text input.
A basic keyboard for a computer is $10 at retail because there is no size constraint making things expensive. Even a $10 keyboard has better user feedback than a touchscreen, but a computer sized touchscreen is going to cost more than $10. Plus, ergonomics. Touchpads could overtake mice, maybe, but desktop is being vastly overtaken by mobile, so it barely matters.
Consumers didn't flock to a more expensive phone (the iPhone in 2007) because it was cheaper to make (it wasn't). They did so because the touchscreen enabled new forms of interaction not yet possible, enabling full-screen games, photo viewing/shooting, and web browsing to name a few.
This type of holographic display will have a similar issue, but it may be saved by the fact that it can be very impressive as a display in commercial uses.
Touchscreen just enabled another method of interacting with web and applications, method that was already present via use of mouses on PC. It did not depend on massive adoption by TV and movie industry with little benefit for them.
No, but I still do use them on my projector. It seems that 3D became pretty popular with the home theater market, who buys more expensive technology than average consumers. This is why most good projectors these days still support 3D, while almost all regular TV's do not.
My, just today there was a piece about SpaceX I watched on the noon news and I noticed that as well. I was amazed, I refused to believe it was true until I looked it up. How could anyone possible think about touch-only interfaces and sci-fi aesthetics in such a critical piece of equipment. It's already preposterous (and dangerous) that they do it on cars, but in a space-ship! that's another level.
Start with medical applications, expand to luxury video chats. Screen is only part of technology - it requires beefy GeForce RTX 2080 TI. Or, looking from another perspective, some potential customers already own $1500 worth of required equipment. Regular 8k displays are no cheap either.
Notebooks and smartphones had a long run before universal success. Lets see in ten years.
What this appears to be is recording pseudo-light felid with thin strips of vertical prisms or lenses with many vertical strips per micro-lens so that you get depth from the horizontal but not vertical perspective (tilting the display up and down won't change the image, but panning left and right will)
https://www.kickstarter.com/projects/lookingglass/the-lookin...
(I recognize the irony of trying to illustrate this with a Youtube video, but nevertheless: https://youtu.be/LkpBYne7SlU?t=54 ; I wish that the one with a man at his desk was viewable.)
This looks to have the same effect, in full color, and animate-able. Light field technology is truly amazing.
Still a fantastic museum to checkout.
That resolution is then divided into the 45 viewing directions: https://docs.lookingglassfactory.com/Appendix/how-it-works/
We need to divide 7680x4320 by the 9x5 grid.
Thus, the effective resolution is only 864x853 scaled in width to make each pixel about 1.75x wider than tall.
At that resolution, it might be difficult to read normal-sized text.
Also, “only 864x853 ... it might be difficult to read normal-sized text.” That makes me feel really old, I spent a long time coding on 640×480 in 16 colors
Every monitor size and resolution has an "optimal" viewing distance where the human eye can resolve the maximal amount of detail. Unfortunately sometimes the "optimal" viewing distance deserves its scare quotes, as it results in the viewer sitting so close that they hurt their neck trying to get away from said overwhelming screen.
For a 55" 4K display, the size of my TV, the optimal viewing distance is 3.7ft, far closer than I sit to my TV. Between 3.7ft and 7.2ft (ideal distance for 1080p), you get some benefit from 4K, but not all of it. In order to even detect the difference for an 8K display I would need to sit somewhere between 3.7ft and 1.7ft (ideal 8K distance) in order to reap the benefits of my purchase. Needless to say, I am not sitting that close.
For monitors the story is a bit different, because they're small and we sit close to them. A 32" 4K display has an optimal range 2ft, which is actually pretty reasonable, while an 32" 8K display has an optimal range of 1.1ft, which is again too close. I personally suspect that this is part of why Apple started pushing to 5K and 6K (although the latter might be cinema related), because ~5-6K is probably the maximum useful resolution for a desktop monitor.
This is a long winded way of saying that 8K is kind of a gimmick, at least for desktop and home use.
Source: https://stari.co/tv-monitor-viewing-distance-calculator
https://xinreality.com/wiki/Vergence-Accommodation_Conflict
Current headsets use two flat screen displays positioned a fixed distance from your head showing two slightly different 2D images. This tricks your brain into thinking you're seeing a 3D environment with objects closer or further away than the displays actually are, but some parts of your visual system are not fooled - leading to a conflict where your eyes try to focus and adjust to what you're seeing in two different ways at the same time.
Holographs may be able to provide more depth cues to each eye, helping to convince the visual system that the images are real.
"""vergence-accommodation conflict, which causes a lot of the motion sickness, headaches, eye strain and other issues people have with VR"""
This is untrue. It actually a pretty minor amount of the issues. There are individuals that suffer from this disproportionately but the estimates I've seen were in the low single digits.
The major issues are in the source to the link you posted -
"""Accommodation-vergence conflict is the one remaining aspect of vision that is not simulated by current VR headsets. While it is not as big a deal as simulator sickness induced by poor tracking, high latency, or artificial locomotion, """
These are the major sources of vr discomfort which are increasingly handled by the baseline vr specification being increasingly in the reach of more and more hardware systems.
I'd also note that as the poor tracking and high latency issues have disappeared people have found they are comfortable with radically wider ranges of artificial locomotion styles.
At this point, in my opinion, the utter uselessness of any text based applications (ie what people actually DO all day long) is what is holding VR back. The resolution needs to scale up fairly radically.
This monitor may have better text results but its not clear what the boundaries of it as a 3d display are as I haven't seen a review from knowledgeable sources (Oliver Kreylos/ docok is one of the people that I'd like to hear from).
That's the biggest culprit. Some people even get nausea without VR headsets. There was a presentation at the California Academy of Sciences where the camera was panning as if it was travelling. A few people nearly barfed.
I wonder how much of it is training. Lots of people are ok with flight simulators, or simulated cars, train rides, etc. It's mostly when they think they are moving that the problem presents itself.
> At this point, in my opinion, the utter uselessness of any text based applications (ie what people actually DO all day long) is what is holding VR back.
In non-gaming scenarios, yes. Resolution could be better. However, it is not so bad at all. You need bigger "displays" in VR than what you would have in real life, but you can code alright.
I would personally prefer lighter, less intrusive devices, even if the resolution was the same.
In vr, taking the camera control away from the user (as if in a film or cutscene) is an absolute no go and whoever did that in the example you experienced should be professionally embarrassed and apologize to the people they put through that.
For animated holograms you have to do this in real time. And colour is still a problem - ideally you want at least three different planes for RGB, all correlated with sub-micron accuracy.
Shortcuts are possible - actually with stereoscopic 3D TVs and monitors they've been and gone - but the real thing won't be happening any time soon.
3D-like emulations - like this product - are much more plausible in the short term.
> When Looking Glass Factory showed /of/off/s its first holographic display way back /on/in/s August 2018, it felt more like a proof of concept than anything — though it was immediately an impressive concept
The live stream of the bird cage is a very interesting application.
Ditto watching the NVDIA keynote recently. They were bragging about how good their AI-driven 8K raytracing was. But I could not see much before and after difference on my puny tablet screen.
The benefits being that you can share the items across many museums at once without exposing them to UV or theft risk, and without taking the originals away from wherever they belong.