171–176 of 176 posts
Earlier quoted context omitted.
But you could build a $30K prototype and show it to investors, saying that the cost would go down when mass production ramps up.
And then not deliver on the pie in the sky promise, or delay long enough that somebody spray solves the issues. Also known as a scam.
Earlier quoted context omitted.
You could approximate this using a standard display if you could dynamically track and respond to focal depth. That might be harder than getting a functional light field display, though.
Much easier, we have the tech in ophthalmology and adaptive lenses that are used in cellphone cameras.
What I was picturing was: - LCD display - Patterned IR image projected onto eye - Camera to read IR image on the back of retina, to figure out current focal length of cornea - Realtime chip to adjust LCD display based on focal depth, to simulate light field image.
... The real issue is that response latency would have to be <10ms probably to avoid being disorienting.
Jeez this sucks. I was always behind Magic Leap with the logic that "big investors like a16z google alibaba wouldn't get suckered" but god, :( this is disappointing.
> I was always behind Magic Leap with the logic that "big investors like a16z google alibaba wouldn't get suckered" Here's a few clarifications that might change your thinking about funding announcements. Google Ventures passed on Magic Leap. Google Capital passed on Magic Leap. Sergey was enamored so Google's own balance sheet provided most of the B round. Word is that A16Z put only $3MM in -- basically enough to ge…
Earlier quoted context omitted.
This is very interesting. Do you have a source on this? I can't find anything online.
Nobody I will quote in a forum(!) but I live in silicon valley where secrecy is poor. If you want to look on the web it's not hard to verify if you're willing to do more than just a google query. For example you can easily tell from the web sites of GV and GC that ML is not in either of their portfolios. I can't find any Form D for magic leap but you can see what some of their investors put in if you dredge some of t…
I thought they had given up a couple of years ago on the gimbal-mounted fiber with the piezo collar. They raised the last round to buy a MEMS fab which I assumed was to do a DLP-like thing down a fibre laminate.
And in fact the Guttag piece linked in this comment page also suggests this (says the DLP demo was the most impressive). I doubt they could use actual DLP (because the TI DLP group is a pain to work with) but it would make sense that their MEMS chip would be also a micro mirror array.
>I presume they're going for something similar to what I'm building in one of the implementations, using switching polarisation-sensitive lenses in a stack.
The other thing they mentioned and which is feasible immediately is using a liquid lens, if the input is small.
This would be used for 45 degree combiner but too small for a waveguide exit or for Oculus size lenses.
---
>Building a single FSD isn't difficult.
Building a single FSD with wide deflection (more pixels) and fast oscillation (more frames and made necessary to counteract the wider deflection) is considerably more difficult.
Building an FSD with the previous attributes an removing jitter/ distortion is more difficult.
Doing all the previous for an assembled array to line up perfectly is more difficult again.
Finally doing an array, particularly a 2D array at scale, and expecting a significant amount of them to actually work is crazy.
The FSD improvements FredzL points to are at odds with each other.
An FSD array can either use to each unit to create a different focus plane/ view or add extra pixels to the 2D resolution for any one instant of time.
One of the University of Washington papers does hint at alternative though.
In the paper it shows a fibre array (line) being placed in a thick block for better alignment.
This would prevent the fibres being vibrated to create the image.
Instead, on a single line, of a different paper they mention trying acousto-optic modulators.
Dennis Smalley had a good paper on the subject, in which ultrasonic waves travelling through a class of materials will locally change the refractive index.
Light bends by different amounts in different refractive indices, thereby allowing the possibility of image formation.
Solid-state scanning arrays or even multiple beams on a MEMS mirror are probably better than FSDs.
They might be able to use FSD, I don't think it's a good idea.
So VR is hard enough - to avoid jitter that makes users feel sick you have to respond to a user's head movement, render a new frame with the new information the user should see, respond to any button presses, then draw the frame, in under 14-20ms. Magic Leap was always a much more difficult problem... they have to respond to a user's head movement, parse the scene the user is looking at (which could be anything), fig…