Slashdot Mirror
John Carmack's Oculus Connect Keynote Probably Had Samsung Cringing
An anonymous reader writes John Carmack, famed keystone developer of 3D networked gaming, has now been working with virtual reality company Oculus for over a year. Much of that time has been spent collaborating with Samsung on the forthcoming Gear VR headset. At his keynote presentation during Oculus Connect, Carmack took to the stage with 90 unscripted minutes of no holds barred discussion of the last 12 months in VR. 'I believe pretty strongly in being very frank and open about flaws and limitations so this is kind of where I go off message a little bit from the standard PR plan and talk very frankly about things,' he said to applause from the audience.
He kept scraping the chalkboard with his Samsung Galaxy S5.
Get free satoshi (Bitcoin) and Dogecoins
On the Gear VR? They've got accelerometer data. It's probably not as precise as external tracking, but still.
It means tracking translation. The Gear VR tracks your head's rotation but not translation which triggers a reflex in your brain that causes nausea.
Basically the brain knows that your head moved a certain way and what your eyes see doesn't match that. In turn, your brain decides that the only explanation is that you've been poisoned and makes you want to hurl as a defense mechanism.
Mod me down, my New Earth Global Warmingist friends!
When was the last time you fingered him?
The aqueduct?
Samsung cringing? Because Carmack referenced hardware limitations of the current display technology that anyone who could follow his speech either already knows or could have gleened from reviewing the basic specs? And the display technology is still is (or is equal too) the best available in industrial quantities?
It's not like he said "Company X's displays are so much better, it's stupid we didn't go with them." That might have induced some cringes. The actual speech? Not so much. It was interesting enough for the technical material, don't try to spoil it with melodramatics.
That was an amazing video purely to watch someone talk without ever finishing a sentence or pausing for thought. It was like watching a stream of consciousness.
Oculus's software uses a ball-stick model of the head, so that turning also produces a translation. Don't knock it till you've tried it (I'm assuming here you haven't tried it), they've almost completely solved the nausea problem, even for sensitive people.
What, Linux distros still run fingerd? That's old geezer Unix crap for dirty greybeard hippies!
You track translation with an accelerometer (like rotation with a gyro). But neither is perfectly repeatable - they don't return exactly to zero. So, without some absolute reference for both, things get off. A compass can provide an absolute reference for horizontal rotation, the accelerometer provides an absolute reference for vertical (tilt - a compass might be able to do both, really), but an absolute reference for translation is more difficult - how do you move across the room and then return to exactly the same place? Apart from external sensors, one might make do with ultrasonic sonar (or infrared lidar?), if willing to live with constraints on the room layout.
"National Security is the chief cause of national insecurity." - Celine's First Law
Ha! Good one. But which server do I finger? We don't have a "planfileserver" infrastructure like we do for gpg keys.
As an aside, Slashdot no longer lets users paste in their public keys into their user information, which is a shame. You can see them at: http://slashdot.org/~username/...
I appreciate how he shares technical details with us even though it probably doesn't benefit him personally at all.
There isn't any study of sufficient size that actually show's they've solved the problem. They have a few anecdotes, and a questionable sample at best. We'll know if they've actually solved the problem when they release it to the mass market, but if they've failed they'll have set VR back to square one.
That isn't actually true. You *will* get sick even with positional tracking, as many people found out when the DK2 Rift was released. Just look in this thread, for example:
https://www.reddit.com/r/oculu... Positional tracking enhances immersion and potentially presence, but it is not really a fix for motion sickness. Unfortunately many people don't understand this.
The problem is deeper - you are correct that the sensory mismatch between what you see and what your sense of balance (inner ear) and proprioception (nerve endings in your muscles relaying the position of your limbs) are telling you is what causes the problem. However, that is not really tied to the positional tracking. It is fairly easy to demonstrate - many people get sick even with full 6DOF tracking using a very expensive big bucks tracking system, walking around in a CAVE, not using an HMD at all (CAVEs are usually far less motion sickness inducing than HMDs).
Most of the nausea problems are caused by poor application design - sudden accelerations are bad, because you don't expect them (it is akin to someone pulling the rug from under you!), motions not initiated by the user are bad (again, unexpected movement!), inappropriate navigation schemes - strafing, head bobbing, "aiming with your head" (not being able to look and change direction of movement independently - as in all FPS games that use mouselook), etc. All these things cause motion sickness. No amount of tracking wizardry is going to help you there unless the design of the application is fixed - and these problems are unfortunately in almost every single demo that was released for the Rift so far, despite there being 30+ years of published research on VR available.
Then there are problems that are often ascribed to motion sickness, but are not really - headaches, dizziness, eye strain. Those are often caused by a poorly adjusted HMD. This is where Rift suffers a lot, because unless you have perfect vision and your eyes are spaced exactly the same as the Rift lenses, you will get eye strain and headache after a while due to a blurry, out of focus image. This is why commercial HMDs have both dioptric adjustment (the two pairs of replaceable lenses really aren't a solution) and interpupillar distance adjustment (the lenses or even displays themselves can be moved closer or farther apart). Another issue with the Rift-like HMDs is with scenes where the textures and jaggy, not antialiased lines cause visible "beating" (moire) against the raster of the relatively low-res display, provoking a lot of visual discomfort - this was really bad in the DK1, DK2 reduced it a bit thanks to the higher resolution and pentile display. That's why dark scenes work best with Rift, because the dark pixel raster is not that visible.
Direct brain interface is the future, but make sure there's a Log Out option once you're inside the game.
Get free satoshi (Bitcoin) and Dogecoins
I hope small amounts of lag in VR causes nausea too, because if there's one thing developers don't seem to care about enough, it's latency/lag and frame rate (no, even 60fps isn't good enough). If they can get it below 10ms lag, then I'll start to be a bit happier.
Why OpalCalc is the best Windows calc
With the money Mr. Carmack earned from the sale of Oculus, do you think he cares? He has suggested that he wouldn't mind simply going back to running his spaceship company, so can Samsung give him a good reason to do that?
This definitely sounds like somebody who doesn't give a damn.
Ubuntu doesn't even install finger client by default!
But OS-X still has it, oddly.
Yep pretty much anything causes it, the VR experience has to be perfect or old brain freaks out.
Crazy things, brains are.
Mod me down, my New Earth Global Warmingist friends!
It's not wrong so much as it's not the complete picture.
But I'm not in the business of writing novels on slashdot.
Mod me down, my New Earth Global Warmingist friends!
Ah I see you've received Oculus DK3.
That one's going to make the billions back for Mr. Zuckerberg.
Mod me down, my New Earth Global Warmingist friends!
Decades ago I had pneumonia and was given floxin as an anti-biotic
It was effective but gave me nausea every time that i went to play wolfenstein 3d, I had to fall back to playing commander keen until I was off the floxin
This is the United States of America. Getting a bone fixed can cost up to $10,000. Imagine the health complications, and doctor bills for problems with a direct brain interface? What if the immune system rejects the interface? What if bacteria colonize the neural interface? What if the body's regular nervous system atrophies from lack of use? What about the human labor in customizing the interface, for each person's unique nervous system?
Compare that to the cost of a 9 inch high 1440p LCD, or even a potential 5760p OLED. Maybe a few thousand dollars at most. Good Human Machine Interfaces will not cause medical problems, and thus no expensive doctor bills. That is why HMI will win.
I can confirm there's an application design issue there: both me and my girlfriend made the mistake of running backwards in the Tuscany demo, and when you do that you can pretty much feel your stomach lurch forwards. There's definitely a learning curve there where VR games are not going to be able to have sudden accelerations like we do with current movement systems. Though conversely, I felt great playing HL2:DM in VR - getting blown about by fans and the like just felt...like well I was being thrown about, but didn't make me feel sick at all.
But I'm not in the business of writing novels on slashdot.
No, because that's Bennett Haselton's job.
Please stand clear of the doors, por favor mantenganse alejado de las puertas
New slogan: "Only OS-X gives you the finger!"
Come play Moral Decay!
Considering I'm sitting here with their latest offering and know you're full of bullshit because I'm not sensitive and the product still makes me feel very unwell in many instances, BULLSHIT
Can I haz your stuff?
This is talking about the Gear though, not the Occulus DK. Has anyone tried the Gear yet?
Carmack? What has he done for us lately?
Carmack is not just riding on his past fame but still actively participating in engineering.
I'm pretty sure you misspelled Romero.
Keep the Classic Slashdot.
This might be the reason it never took off the fist time around, 15 years ago, and why it's still failing to take off. For VR to work, you have to design the game specifically to work with the VR hardware. But the VR hardware has always been so expensive that only a few people are willing to buy it, and therefore there is not incentive to spend the time developing for it. And since there's no development going on, there's very little reason for people to spend so much money on it. Big circular problem that's hard to fix. Really they only way I can see this coming to an end is for the hardware to get really cheap ($200), and then have some huge powerhouse behind it who can push it and get great games on the shelves at release time.
Anthropic principle: We see the universe the way it is because if it were different we would not be here to see it.
It means tracking translation. The Gear VR tracks your head's rotation but not translation which triggers a reflex in your brain that causes nausea.
Basically the brain knows that your head moved a certain way and what your eyes see doesn't match that. In turn, your brain decides that the only explanation is that you've been poisoned and makes you want to hurl as a defense mechanism.
Maybe not such a bad thing. A lot of gamers drink Monster, which is similar to poison - at least in flavor.
This isn't about frames per second but latency of each frame.
Today most GPUs render things into a buffer, and render two-three frames ahead (double or triple buffering). Far more important, as noted by Carmack in the speech (seriously, go watch it, it's very informative) is being able to do just-in-time rendering and send it straight to the framebuffer to minimize latency.
I did watch it. I was pleasantly surprised to see a focus on latency and fps throughout.
Perhaps G-Sync and a high frame rate is a simpler way to reduce latency, and then we can still keep the three frame pipeline. I don't think he mentioned G-Sync which was a surprise. JIT rendering also sounds good though.
Why OpalCalc is the best Windows calc
G-sync is about syncing output of the buffer to frame rate to avoid tearing. It does nothing to latency.
JIT rendering also made it to maxwell btw, but that's a whole different story.
http://www.geforce.com/whats-n...
Gsync does help with input lag / latency. From the horse's mouth, it's mentioned numerous times: http://www.geforce.com/hardwar...
Why OpalCalc is the best Windows calc
The only way it could possibly "help latency", other than placebo effect, is helping latency vs system with vsync enabled.
It cannot do anything to the system with vsync disabled, and having to match frame timing on display likely increases latency overhead. Monitor overhead already occupies a large chunk of the latency, and increasing it would likely increase, rather than decrease input latency.
If you were conscious of input latency, you were already playing with vsync off. That means gsync will in fact increase your input latency if enabled by increasing latency overhead on the monitor, just to a lesser degree than vsync.
The article in the link you provide as evidence agrees.