Hacker News Reader

I think you're missing the bigger point that he's trying to make - our sensory systems work in concert to provide a full picture of reality and they're very sensitive and extremely well tuned together through a lifetime of experiences.

Just providing an accurate projection of the visual field to the eye's doesn't solve the problem completely. Visual perception is not a passive mechanism, our eyes 'explore' the visual field through quick movements and readjustments and this process inherently locks together the proprioceptive signals from the ciliary and ocular muscles to the stream of information from the retina. By only providing screens, we can't simulate this process in a natural way and our body will detect it.

Similarly - our vestibular sense is strongly tied to our experiences of our body moving through the world. Just using a blanket stimulation of electrodes on the skin will not recreate these signals in anywhere near a convincing enough manner to fool the brain's perceptual processing.

All of our sensory perceptions are strongly linked together - if just one is perturbed in an unconvincing way (try wearing glasses that aren't prescribed to you for a while), our brain will do our "caveman brain" thing and make us sick.

> our eyes 'explore' the visual field through quick movements and readjustments and this process inherently locks together the proprioceptive signals from the ciliary and ocular muscles to the stream of information from the retina. By only providing screens, we can't simulate this process in a natural way and our body will detect it.

That's the point of the light field displays GP mentioned. They also encode depth allowing the eyes to refocus on the correct depth corresponding to the object's virtual position.

Put differently, a LFD reconstructs photon paths, as if you were looking into a window facing the recorded object.

The remaining engineering problem is that those microlens arrays step down the effective resolution of the display, so you need much higher DPI to get the same sharpness as a non-LFD would get you.

On the other hand the brain is insanely adaptable. People made to wear glasses that flipped their vision upside down adjusted fine, and when they took them off, everything looked upside down.

Even your example is wrong. You can in fact adjust to glasses of the wrong prescription. People who need glasses have deformed eyes which is basically like having the wrong prescription. The brain adjusts, and people can go without glasses without vomiting everywhere.

The nausea reaction is an issue, but it doesn't affect the entire population. I have friends who can't play video games because of motion sickness, but no one argues that video games can't be commercially successful. And various tricks like removing head bob, adjusting the field of view, and adding a virtual "nose", have been shown to significantly alleviate that.

>They're comparing apples to oranges. The $80,000 military VR headsets of 1989 absolutely did not use OLED or OLED light field displays.

Who told you he refers to 1989 headsets only? The military also has 2016 $80,000 (and more) headsets.

The important costs though for the current wave of VR are hundreds of millions of dollars for OLED display manufacturing lines, and an additional many millions of dollars of engineer time from Nvidia / AMD / HTC etc for custom drivers, low persistence hardware etc. It's very, very, likely that previous $80,000 or even $200,000 systems are not up to par with current mass-market hardware regarding key metrics relevant to motion-sickness, or even that the military bidding process would even specify those metrics considering that their importance was only discovered recently through user testing on hardware that was not previously available.

One other interesting thing to note (at least on the HTC Vive) is the majority of the content for the platform is focused on standing / walking with motion in real life matching 1-1 with motion in VR. Which both stops a major source of motion sickness and rules out entire categories of software like flight simulators, which seems to be his entire work history.

Computer software and hardware are an interesting case where often, the more expensive something is, the worse it is; the best stuff is often quite cheap. This comes from the high upfront costs meeting very low marginal costs. A cutting-edge chip fab will cost $10b+ all told right now, but can produce millions of chips a year at trivial costs.

I knew as soon as he mentioned military VR results that his case was going to be based on ancient and inferior hardware and outdated results, and include no actual results or even first-hand experience with a Vive. (At the most generous reading of the article, he worked with a DK1 a while ago.)

1-to-1 motion certainly seems to eliminate motion sickness. I can't remember a single instance of getting nauseous with a room-scale Vive experience and I've been in that headset for numerous hours.

I've tried a couple seated flight-sims. You simply can't see distant objects as well as you'd expect due to the low resolution of the panels, which is far more important in flight sims than with most other things. I can't say I got sick, but it didn't feel real to me, and I can see how others might get sick.

4K and, later, 8K VR should solve a lot of that.

> vestibular stimulation

Whoa! I have been kind of watching the whole VR thing from the sidelines, and I didn't expect this to be even anything somebody was even thinking about.

Thanks for the hint! Are there any specific projects that you'd think are worth mentioning here ?

Entrim 4D (out of Samsung's C-Labs) seems to be a bigger one.

There are consumer oriented research GVS headsets currently under development, Samsung had one at CES this year. From what've heard it's not hugely accurate, but the general sensation should be enough.

Light-field displays have not solved anything. They introduced depth at the expense of resolution, and by a huge factor. Even regular VR headsets have huge pixels, the light-field ones are so low resolution, you will wish your 320x240 VGA was back.

If current progress with MicroOLED panels continues, we should expect 5000x5000 displays that are small enough for use in a glasses form factor with light field displays. This will significantly help with the reduced effective resolution.

The next problem will be figuring out how to render 10000x5000 at 120hz in realtime on consumer GPUs. It will definitely require insanely beefy rigs to accomplish.

10000x5000 displays will take decades. That's a factor of 6 compared to 4K. We don't even have graphic cards that can play modern games at 4K / 120Hz, we just got there for 1080p.

> we just got there for 1080p

I've been playing modern games at 1440p 120hz since 2011. We've been at 4K and 5K 120hz-capable (on the GPU-side) for quite a while already as well.

It's hard to imagine not being able to hit >8K 120hz in under 500W once we get GPUs on the 7nm node.

Of course the actual displays aren't here yet, but the compute resources needed aren't out of reach.

I'm talking about modern games, meaning 2016, not 2011. Maybe you can play some lame (graphics wise) game at 4K 120Hz, but not the ones that are actually huge.

I literally said modern games, meaning whatever is the most modern at the time, since 2011, not only at 2011.

You don't necessarily need to render everything at full resolution, just the area the eyes are focused on which you detect with good enough eye tracking.

"galvanic vestibular stimulation"

WOW interesting: https://www.youtube.com/watch?v=OlXYqfQHNuA