4 ms·

Supplementary: The 3D Printed Occulus Rift killer. http://www.durovis.com/index.html http://www.durovis.com/index.html

by X4 13y ago

georgemcbay 13y ago
Good luck with that. What makes the Occulus Rift important is how fast it translates real world movement to signals software can process to give a true VR feel. The accelerometers and magnetometers in phones are decent for their intended purpose, but their intended purpose is not VR. This is no Rift killer.
- X4 13y ago
  That's speculation. Do you have factual evidence to prove that current accelerometers aren't responsive enought? Like a datasheet, a publication or something similar that isn't based on gut feeling. No offense, I appreciate the Occulus Rift as a high tech device that sets new limits in VR Technology, but Smartphones have come a long way sir. Are you saying that the iPhone4s, iPhone5, Galaxy S3, Galaxy S4 and Google Nexus phones accelerometers are really unresponsive? Btw. just in case your theory is right (I admit when I'm wrong), then please checkout how long that theory will have legs ;) * Quantum Limit, freaking yeah!! http://arstechnica.com/science/2012/12/an-accelerometer-so-good-that-only-quantum-mechanics-limits-it/ http://arstechnica.com/science/2012/12/an-accelerometer-so-g...
  - acgourley 13y ago
    See my sibling comment to yours for more context. There are two main problems with the sensors in current devices. The first is that the communication layers (hardware and software) between the sensors and your applications can add 60ms or more. The rendering loop itself will add another 16ms (60fps) or 33ms (30fps). The second problem is that they fast, but not very fast. Generally phone position sensors can run 100 times a second. The OR sensors, on the other hand, run at something like 1000 times a second. This lets a processor in the goggles themselves do predictive algorithms which guess at future head motion allowing the higher level software to compensate for expected head motion. The key is that humans can perceive latency around 20-30ms, and the phone is delivering around 80-100 in the best case. For what it's worth, I did real measurements on all this before giving up on the idea myself.
    - X4 13y ago
      hmm.. I know about the humans limit, that's important to begin with. I don't know details about the latency. My busted post doesn't have that backing, but the sensitivity of 16G is enough for VR applications.
    - X4 13y ago
      I just looked through the Datasheet and I have to correct my post. It HAS DATA about the latency, but I don't fully understand it. It says something about 2000 times a second. Please open the PDF on page 14, chapter 2 Module specifications.
      - Someone 13y ago
        Latency is measured in seconds, not Hz. I can film the sun at 2, 2 thousand or 2 million images a second, but when I do that from earth, the latency will never (1) be below 8 minutes. (1) nerd remarks: yes, it will eventually go below it, when the sun expands to become a red giant. Also, earth might be knocked out of its orbit earlier, or maybe we move it, etc.
- acgourley 13y ago
  I had the opportunity to ask Palmer Luckey (the founder of OR) directly about the possibility of smartphone based HMDs. It turns out he's looked into it quite a bit, but ultimately the unsolvable latency of the sensors made him move on. It could work for viewing 3D content or for situations where the head will only be moved fairly slowly, but it's not a great solution for most applications. I would note that a dedicated handset maker could create the hardware and drivers which fix this.
  - X4 13y ago
    I highly appreciate your feedback. Especially that the founders looked into that possiblilty, just shows how much class they have. Carmack doesn't support them just because they're cool. There is more to it.
- X4 13y ago
  OK, I just busted your theory. Sorry. :) It has 16G The LSM330 has an user-selectable full scale acceleration range of ±2 g/±4 g/±6 g/±8 g/±16 g and angular rate range of ±250/±500/±2000 d. Samsung Galaxy S4 accelerometers/gyro Datasheet: http://www.st.com/web/en/catalog/sense_power/FM89/SC1448/PF253882 http://www.st.com/web/en/catalog/sense_power/FM89/SC1448/PF2... Here's a FULL Teardown of the insides of the Samsung Galaxy S4. http://www.chipworks.com/blog/recentteardowns/2013/04/25/inside-the-samsung-galaxy-s4/ http://www.chipworks.com/blog/recentteardowns/2013/04/25/ins...
  - georgemcbay 13y ago
    Even if there were a single phone that outperformed the Rift in real world responsiveness (and there isn't), that wouldn't really change anything because the whole value proposition of a generic smartphone printed solution is you use it with whatever phone you already have. If you have some phone other than that super phone, as the majority of smartphone users would, you then have to buy some specific $650 phone for the thing to give you comparable performance to the Rift, and you just spent more than double what a Rift costs.
    - X4 13y ago
      Who said that your ordinary iPhone4s or iPhone5 hasn't comparable, equal or better sensors? Am not here to speculate and I know it's tedious to prove everything, but I welcome it in your next post :) Everyone benefits, if one bright mind comes up with a solution or a fact that others have to live with.
  - snogglethorpe 13y ago
    Note that the attributes of the phone hardware sensors are important, but not sufficient—generic "use any phone, no root required!" solutions like you mentioned will have to send the data through a full stack of software layers that are very unlikely to be optimized for high-rate / low-latency uses like this one. You might get lucky and have a phone so incredibly powerful that even non-optimal software stacks are fast/low-latency enough to work well, but I suspect it will be a long time before most people do...

Supplementary: The 3D Printed Occulus Rift killer. http://www.durovis.com/index.html http://www.durovis.com/index.html

by X4 13y ago

Supplementary: The 3D Printed Occulus Rift killer. http://www.durovis.com/index.html http://www.durovis.com/index.html

Good luck with that. What makes the Occulus Rift important is how fast it translates real world movement to signals software can process to give a true VR feel. The accelerometers and magnetometers in phones are decent for their intended purpose, but their intended purpose is not VR. This is no Rift killer.

X4 13y ago
That's speculation. Do you have factual evidence to prove that current accelerometers aren't responsive enought? Like a datasheet, a publication or something similar that isn't based on gut feeling. No offense, I appreciate the Occulus Rift as a high tech device that sets new limits in VR Technology, but Smartphones have come a long way sir. Are you saying that the iPhone4s, iPhone5, Galaxy S3, Galaxy S4 and Google Nexus phones accelerometers are really unresponsive? Btw. just in case your theory is right (I admit when I'm wrong), then please checkout how long that theory will have legs ;) * Quantum Limit, freaking yeah!! http://arstechnica.com/science/2012/12/an-accelerometer-so-good-that-only-quantum-mechanics-limits-it/ http://arstechnica.com/science/2012/12/an-accelerometer-so-g...
- acgourley 13y ago
  See my sibling comment to yours for more context. There are two main problems with the sensors in current devices. The first is that the communication layers (hardware and software) between the sensors and your applications can add 60ms or more. The rendering loop itself will add another 16ms (60fps) or 33ms (30fps). The second problem is that they fast, but not very fast. Generally phone position sensors can run 100 times a second. The OR sensors, on the other hand, run at something like 1000 times a second. This lets a processor in the goggles themselves do predictive algorithms which guess at future head motion allowing the higher level software to compensate for expected head motion. The key is that humans can perceive latency around 20-30ms, and the phone is delivering around 80-100 in the best case. For what it's worth, I did real measurements on all this before giving up on the idea myself.
  - X4 13y ago
    hmm.. I know about the humans limit, that's important to begin with. I don't know details about the latency. My busted post doesn't have that backing, but the sensitivity of 16G is enough for VR applications.
  - X4 13y ago
    I just looked through the Datasheet and I have to correct my post. It HAS DATA about the latency, but I don't fully understand it. It says something about 2000 times a second. Please open the PDF on page 14, chapter 2 Module specifications.
    - Someone 13y ago
      Latency is measured in seconds, not Hz. I can film the sun at 2, 2 thousand or 2 million images a second, but when I do that from earth, the latency will never (1) be below 8 minutes. (1) nerd remarks: yes, it will eventually go below it, when the sun expands to become a red giant. Also, earth might be knocked out of its orbit earlier, or maybe we move it, etc.
acgourley 13y ago
I had the opportunity to ask Palmer Luckey (the founder of OR) directly about the possibility of smartphone based HMDs. It turns out he's looked into it quite a bit, but ultimately the unsolvable latency of the sensors made him move on. It could work for viewing 3D content or for situations where the head will only be moved fairly slowly, but it's not a great solution for most applications. I would note that a dedicated handset maker could create the hardware and drivers which fix this.
- X4 13y ago
  I highly appreciate your feedback. Especially that the founders looked into that possiblilty, just shows how much class they have. Carmack doesn't support them just because they're cool. There is more to it.
X4 13y ago
OK, I just busted your theory. Sorry. :) It has 16G The LSM330 has an user-selectable full scale acceleration range of ±2 g/±4 g/±6 g/±8 g/±16 g and angular rate range of ±250/±500/±2000 d. Samsung Galaxy S4 accelerometers/gyro Datasheet: http://www.st.com/web/en/catalog/sense_power/FM89/SC1448/PF253882 http://www.st.com/web/en/catalog/sense_power/FM89/SC1448/PF2... Here's a FULL Teardown of the insides of the Samsung Galaxy S4. http://www.chipworks.com/blog/recentteardowns/2013/04/25/inside-the-samsung-galaxy-s4/ http://www.chipworks.com/blog/recentteardowns/2013/04/25/ins...
- georgemcbay 13y ago
  Even if there were a single phone that outperformed the Rift in real world responsiveness (and there isn't), that wouldn't really change anything because the whole value proposition of a generic smartphone printed solution is you use it with whatever phone you already have. If you have some phone other than that super phone, as the majority of smartphone users would, you then have to buy some specific $650 phone for the thing to give you comparable performance to the Rift, and you just spent more than double what a Rift costs.
  - X4 13y ago
    Who said that your ordinary iPhone4s or iPhone5 hasn't comparable, equal or better sensors? Am not here to speculate and I know it's tedious to prove everything, but I welcome it in your next post :) Everyone benefits, if one bright mind comes up with a solution or a fact that others have to live with.
- snogglethorpe 13y ago
  Note that the attributes of the phone hardware sensors are important, but not sufficient—generic "use any phone, no root required!" solutions like you mentioned will have to send the data through a full stack of software layers that are very unlikely to be optimized for high-rate / low-latency uses like this one. You might get lucky and have a phone so incredibly powerful that even non-optimal software stacks are fast/low-latency enough to work well, but I suspect it will be a long time before most people do...