Jump to content
The Dark Mod Forums
Sign in to follow this  
LDAsh

John Carmack does a JRE

Recommended Posts

Carmack is SO awesome, man! :D


Task is not so much to see what no one has yet seen but to think what nobody has yet thought about that which everybody see. - E.S.

Share this post


Link to post
Share on other sites

It was a good interview, he's a lot better at talking to less technical people now.  Interesting perspective on why he's interested in VR.  You can tell that he gets shit for not transitioning to a purely managerial position but I respect that he's adamant about sticking to his passions.

Little disappointed that he left Joe with the impression that you can't really do smooth locomotion in VR.  I get that Carmack is basically only concerned about bringing VR to "the masses", but more intense gamers (e.g. the type that would play Quake competitively) will build up to it.  In my experience it takes most gamers a about a week with daily exposure of 10 to 30 minutes to get translational movement down, but you need to be consistent and careful to not push until you're actually sick--which is the actual hard part because most casuals are impatient and not very dedicated.

However I have to wonder if this push to bring VR to the masses will fail because I'm not so confident that the masses, casual gamers, etc etc that Carmack wants will be interested in VR for quite some time.  First there are all of its current discomforts--visual, ergonomics, and sim sickness.  Second, it currently suffers from certain critical limitations, e.g. fears of sim sickness lead developers to dramatically reduce player agency, the lack of variable focus makes it difficult to interact with things in the near field, and the lack of positional/rotational haptic feedback makes certain actions clunky.  Lastly, it isolates you from your surrounding environment and many applications require a decently sized space and high energy input.  All of these things scream to me "still high end/enthusiast".  Rather, I think the more realistic scenario is that VR continues in the high end and gradually expands as problems like the above are addressed.  I just doubt that FB has the patience for that--I expect them to transition to pure AR at some point because that is *certainly* a technology for the masses even in a limited form (as long as it doesn't make you look like a borg)

 

Lastly, this whole focus on standalone VR is pretty absurd to me.  I mean, yes Carmack is right that Quest is an incredible device--it is technically quite amazing (I've used it)--but wireless and a dedicated external processing box will be just as convenient.  In addition, VR is dramatically increasing processing loads, and not just for stereoscopic, high fov, high resolution, high fps graphics processing but also with respect to physics and interaction--when you give players less abstract and higher fidelity inputs the simulation itself must also expand and converge on something higher fidelity.  I'm not sure how Carmack squares the dramatic increase in processing demands with something that is, as he says, ~1/50th of the power of desktop PCs--which are already struggling.  I mean, I consider myself a hardcore VR enthusiast with rock solid VR legs--I'm regularly zipping around at 800km/h around tight turns in BallisticNG with its VR cockpit view, but the Quest's 72hz display made me feel pretty bad (not to mention it's often motion smoothing from a lower fps up to 72).  A traditional console with a flat mode makes way more sense for the reasons noted above, and also in terms of price: people are going to buy the next iteration of their console anyway--independent of any VR-related desires--as very few people are willing to completely replace flat gaming with VR right now, and thus there is no cost advantage to standalone for most everyone.

It's just anecdotal but I bought my brother a Quest a few months back.  He tried using it maybe 8 times but ultimately put it back in the box.  At the end of the day he's tired and doesn't want to wear something fatiguing.  Right now I'm hoping he might use it for exercise--I think that's the only salvageable use case for him (which was my original intention anyway)

Edited by woah

Share this post


Link to post
Share on other sites
39 minutes ago, woah said:

the lack of variable focus makes it difficult to interact with things in the near field

There are a number of ways to handle this and Oculus is working on it.

https://www.roadtovr.com/facebook-oculus-half-dome-prototype-vr-headset-140-degree-varifocal-f8/

Ultimately, lenses could become "flat" metalenses or be replaced by lasers beamed into the retina, allowing VR or AR in a form factor comparable to a pair of sunglasses. This will lower the weight of the "headset" and make it more appealing to consumers.

Standalone MUST be the goal. That broadens the market, would let you use it while traveling, let you run around a large building or field, etc. And it reduces latency and the need for an ultra high speed line of sight wireless connection or tether. 16K resolution per eye and 240 Hz to 1000 Hz is the goal, and yes, smartphone SoCs will become capable of delivering the necessary petaflops-scale performance at some point. We will see 100x and possibly 1,000,000x performance improvement of chips in the coming decades using some new transistor technologies and 3D chips with RAM or universal memory physically touching the cores. I will go out on a limb and say that 3D NAND or universal memory in a microSD/UFS/3DSoC/etc. form factor will be able to put at least 10 terabytes of storage in a headset for games and applications.

Edited by jaxa

Share this post


Link to post
Share on other sites

It's easy to transition from working behind a computer to playing a game on the same computer. Likewise it's easy to transition from consuming media on a phone to playing a game on that phone. Although I think Quest is the right step forward, mobile VR is the best way forward. And even then, I think there are a few more problems.

Something like Google Cardboard, Daydream or Gear VR are still separate devices just like Quest. A mobile phone would need easily hidden unfoldable VR flaps so anyone with just a phone could easily transition to VR content without having to take such content into account in advance and bring something along for that occasion. Finally, VR is both isolating and any VR goggles don't look styling in public. So no one will do VR in public even if were possible and easily accessible, while mobile gaming in public is not a problem.

AR has much more potential as long as the goggles look stylish, like sunglasses. Nobody minds wearing those in public. And if they provide a friendly overlay showing highly rated places and a route to them, I can see that becoming very popular. It's not styling, reliable and cheap enough yet, but this is a step forward for AR:

 

Share this post


Link to post
Share on other sites
1 hour ago, jaxa said:

There are a number of ways to handle this and Oculus is working on it.

https://www.roadtovr.com/facebook-oculus-half-dome-prototype-vr-headset-140-degree-varifocal-f8/

Ultimately, lenses could become "flat" metalenses or be replaced by lasers beamed into the retina, allowing VR or AR in a form factor comparable to a pair of sunglasses. This will lower the weight of the "headset" and make it more appealing to consumers.

Standalone MUST be the goal. That broadens the market, would let you use it while traveling, let you run around a large building or field, etc. And it reduces latency and the need for an ultra high speed line of sight wireless connection or tether. 16K resolution per eye and 240 Hz to 1000 Hz is the goal, and yes, smartphone SoCs will become capable of delivering the necessary petaflops-scale performance at some point. We will see 100x and possibly 1,000,000x performance improvement of chips in the coming decades using some new transistor technologies and 3D chips with RAM or universal memory physically touching the cores. I will go out on a limb and say that 3D NAND or universal memory in a microSD/UFS/3DSoC/etc. form factor will be able to put at least 10 terabytes of storage in a headset for games and applications.

I'm aware of that research but I just think this is going to take a long time.  I think we will be really lucky if we get variable focus by the time second generation headsets come around ~2024.  My stance on things is that we'll be fortunate if the high end enthusiast market sustains itself until that time--I think variable focus will breach a certain threshold of comfort and immersion that will make the technology desireable as a pure substitute for monitors in traditional gamepad/M&K gaming (still with snap rotation) and some desktop applications.  I'm also really hoping that anti-simsickness tech like this pans out https://www.defenseone.com/technology/2018/11/inventor-may-have-cured-motion-sickness-without-drugs-and-could-mean-lot-us-military/152960/  .  I love VR and I use a Valve Index every day.  I am not dismissing the technology overall--just trying to be realistic.  New technologies like this almost always start in the high end and gradually expand to wider markets as both the hardware and software experience improves and is shaped by the experience of early adopters.

The real question is whether or not short term profit obsessed corporations like FB, Google, MS, and Sony will have the patience for that.  MS is already pivoting to enterprise and there is no sign of VR for their next console.  Google Daydream is practically dead and they seem to have largely lost interest in the tech--I suspect they're going to focus on AR (and they should).  Sony's public comments and leaked patents seem to suggest that they might try to stick it out--but again they are smart enough to not release a pure VR platform.   A leaked email from a few years ago reveals that Zuckerberg expected he would have tens of millions of users right now--in retrospect that is laughable.  What I see FB doing is waiting 2 to 3 years after an absurdly overhyped first generation of hardware and then getting impatient and launching mainstream consumer focused and aggressively subsidized products when the technology is too early for that.  Hence the leaks late last year about the whole "race to the bottom" thing and the Rift 2 being cancelled, I think not coincidentally timed with the departure of the Oculus co-founder Brendan Iribe.

 

I still don't understand why standalone must be the goal for VR.  It's solving problems that don't exist.  Aside from enterprise/business applications and frequent travelers, how many consumers are actually going to regularly use VR outside of a dedicated playspace in their home?  It's enough of a pain in the ass to just carry a cellphone around with you, let alone take it out of your pocket, occupy your attention with it, and fumble around with a crappy touchscreen UI whenever you get pinged.  But carrying around a full VR headset?  That people running around in large open fields or warehouses with VR headsets on will be anything but a very niche or occassional OOHE application sounds nuts to me.    On the other hand, you throw on e.g. a PSVR2 with inside-out tracking, it wakes up your console from sleep within moments, and the 802.11ay wireless (which will have negligible latency) gives you all of the power of a PS5 ... that you would've bought anyway even if you weren't interested in VR.  And that PSVR2 would not have the costs of shoehorning decent processing into such a small device--it would likely cost less than or equal to the cost of the standalone device with much better specs.

As for advances in processing giving standalone the edge, while I am skeptical of Carmack's understanding of people, their tolerances and what they desire, I do trust his outlook on technology and he stated in this very interview that he thinks fundamental limits to the performance of silicon will bottleneck us well before mobile chipsets get anywhere near today's desktop performance.  And again, we're already struggling with VR here.  Eyetracking and foveated rendering will help but we're still looking at a dramatic increase in pixel density, FOV, and physics processing loads.

On the other hand, I think AR makes complete sense even in a limited rendition if they can get the form factor small enough.  Just being able to do basic things like go to the grocery store and nearly effortlessly and seamlessly identify and check off items would be transformational (vs the absurdity of trying to do this via a clunky smartphone app and UI). Nevermind all of the obvious media, entertainment, and communication based applications of AR.  It fits the business model of the aforementioned corporations to an absolutely frightening extent--AR will be something you wear nearly every waking hour and they will have moment to moment access to everything you're doing.  There are still a lot of technical challenges here, such as occlusion, but this is the holy grail for them (short of some BCI-based interface).  It will be interesting if they can resist the urge to make everyone look like a branded tool or borg--I think this will be a major selling point for the early versions.

Edited by woah

Share this post


Link to post
Share on other sites

Well, I forgot about foveated rendering. If that's applied and it can reduce bitrate requirements by 90-95%, maybe 40 Gbps 802.11ay is fine, even for 16K/240Hz. My estimate is that 1.28 Tbps is needed for 16K/240 Hz, 4:4:4, 10-bit color depth. So you may still need display stream compression to get under 40 Gbps.

The bulk of headsets will go down, possibly all the way down to a pair of sunglasses that you put a cover on to switch from AR to VR mode.

The smartphone-based VR headset is pretty reasonable, since most people constantly carry a smartphone. So you just need to remember to throw a headset into your bag. There are smartphones with 4K screens, 120-144 Hz, and any additional resolution would only be beneficial for VR at this point. I'd say the failure of Daydream is more of an indictment of slow adoption of VR in general. There needs to be lots of 180-360-degree video content, more games and applications, and hardware/lens improvements.

One thing you should consider are driverless cars, which should see some uptake by 2030. If you have 1-4 hours of commute every day, that time could be spent using a VR headset instead of driving. Or maybe that will make you queasy. Otherwise, you will have to suffer from in-vehicle advertisements, unless you own the driverless car.

Edited by jaxa

Share this post


Link to post
Share on other sites

Join the conversation

You can post now and register later. If you have an account, sign in now to post with your account.

Guest
Reply to this topic...

×   Pasted as rich text.   Paste as plain text instead

  Only 75 emoji are allowed.

×   Your link has been automatically embedded.   Display as a link instead

×   Your previous content has been restored.   Clear editor

×   You cannot paste images directly. Upload or insert images from URL.

Sign in to follow this  

×
×
  • Create New...