cancel
Showing results for 
Search instead for 
Did you mean: 

Foveated rendering seems unlikely for CV2

Rob_In_Phoenix
Rising Star
I happened upon this youtube link in the Oculus Subreddit:

https://www.youtube.com/watch?v=2EfOJOahTOQ&feature=youtu.be

At the 10:00 mark, Palmer answers a question about foveated rendering.  It doesn't seem like eye tracking is ready for prime time, I have my doubts this will be in the CV2.

I don't know about you all, but this was kind of news to me.


13 REPLIES 13

Protocol7
Heroic Explorer
From last year's presentation at OC3, Michael Abrash makes predictions about VR in 5 years time (so 4 years now). He sees reliable eye tracking as an absolutely crucial but extremely difficult element.

(At 14m53s he talks about eye tracking.)
https://www.youtube.com/watch?v=AtyE5qOB4gw&t=14m53s

Great eye tracking is so central to the future of VR that I believe it will be solved 5 years from now. Although I have to admit it is the greatest single risk factor for my predictions.

It would be fair to assume that Oculus and its competitors are all working hard trying to solve this problem




Zenbane
MVP
MVP
Hmm, but wasn't eye tracking one of the Achievement Unlocks from the Pimax 8K Kickstarter stretch goals?
😮

Facebook owns an eye-tracking company (The Eye Tribe). I wouldn't worry about it too much. Besides,  citing Palmer as a source is bad for ones mental health lol

DarkTenka
Trustee
To clarify, he says hes pessimistic about "near future" applications of Eye Tracking, due to the problem of consistent calibration both with multiple different faces/eyes/pupils and with possible minor shifts with the physical headset.

He also says he believes that it is a definite thing for the long term and that it's definitely the way to go .. just that it poses a significant technical challenge with current hardware.

You can interpret that many ways.. if "Long Term" is 2-3 years then I am still hopeful for CV2 w/ Eye Tracking. I believe his current "pessimism" with Eye Tracking is more, rather subtly, aimed at Pimax and Fove.

Anonymous
Not applicable
Meh, but I still consider any headset with good working eye tracking and higher than current gen resolution a real VR2 over just res and FOV increase. I know it might not be a thing - but I think they should really focuses on it up until release to see if they can make it work at least 80% of the time. I know 20% is a large number to still over come - but it's something that can be work on software and best fix for software can still be a big win over all than not having it at all.

Anonymous
Not applicable

Zenbane said:

Hmm, but wasn't eye tracking one of the Achievement Unlocks from the Pimax 8K Kickstarter stretch goals?
😮

Facebook owns an eye-tracking company (The Eye Tribe). I wouldn't worry about it too much. Besides,  citing Palmer as a source is bad for ones mental health lol



Eye tracking != foveated rendering.

The eye tracking that the Pimax team are using is purely to track eyes to make avatars behave more realistically.

kzintzi
Trustee

snowdog said:


Zenbane said:

Hmm, but wasn't eye tracking one of the Achievement Unlocks from the Pimax 8K Kickstarter stretch goals?
😮

Facebook owns an eye-tracking company (The Eye Tribe). I wouldn't worry about it too much. Besides,  citing Palmer as a source is bad for ones mental health lol



Eye tracking != foveated rendering.

The eye tracking that the Pimax team are using is purely to track eyes to make avatars behave more realistically.


it's a bit of a gimmick, but if it then lends itself to allowing FOVeated rendering down the track then cool.
Though you are more than slightly incoherent, I agree with you Madam,
a plum is a terrible thing to do to a nostril.

shiari
Heroic Explorer

kzintzi said:


snowdog said:


Zenbane said:

Hmm, but wasn't eye tracking one of the Achievement Unlocks from the Pimax 8K Kickstarter stretch goals?
😮

Facebook owns an eye-tracking company (The Eye Tribe). I wouldn't worry about it too much. Besides,  citing Palmer as a source is bad for ones mental health lol



Eye tracking != foveated rendering.

The eye tracking that the Pimax team are using is purely to track eyes to make avatars behave more realistically.


it's a bit of a gimmick, but if it then lends itself to allowing FOVeated rendering down the track then cool.

Entirely different thing. Displaying some eyes on an avatar you can get away with being inaccurate, uncalibrated, and having some latency. With foveated rendering all that changes: it has to be precise and extremely low latency (your eyes are very, very quick, probably even would need to be predictive).

People may be hearing "foveated rendering" when they read "eye tracking" but it's really not the same thing.

Fri13
Protege



Entirely different thing. Displaying some eyes on an avatar you can get away with being inaccurate, uncalibrated, and having some latency. With foveated rendering all that changes: it has to be precise and extremely low latency (your eyes are very, very quick, probably even would need to be predictive).

People may be hearing "foveated rendering" when they read "eye tracking" but it's really not the same thing.


I watched a interview of the one engineers for some other VR company about foveon rendering and he explained about the human eye speed to react and adjust something after movement, that it is actually a very slow. 
This is one of the reasons why the humans has about 1 second reaction time for a new elements as the eyes doesn't focus fast enough for the new situation, and why the human reaction times at the best averages above 200-250ms as the change is actually spotted with such delay. 

This was like the good thing for the eye tracking technology as after detecting the eye movement, the foveon rendering system has time to actually increase details in the new area.

Personally I have three wishes that I want to see in CV2 in 2020 (I don't think it should come earlier, nor expect it to come so current ones would get things working with the current CV1 and Vive):

1) Foveon rendering
Just so we can lower the hardware requirements and get the VR more functional for older (current) computers. This is actually reason why I would want to see such HMD out ASAP, if we could get a 30-50% rendering performance increase.

2) Increase in the resolution.
I want to see 5K HMD (5120 x 2880) so one eye would get a 2560x1440 resolution panel, instead current 1080x1200. It would as well allow a wider FOV in horizontal. 

3) Slightly wider FOV for horizontal use. With current CV1 I don't mind the black bars top or below but the narrow view from sides gets sometimes break the immersion. Just get it slightly wider so it is more like looking through sun/eye glasses. It really ain't big difference to current one. And one possible nice trick to make "immersion" larger could be using a tiny LED lights around the lenses to lit up in colors that is the edge of screen image. More like some of the Phillips (?) televisions has the LED behind display to change television surrounding colors to match the movie so it is more immersive. Hack with arduino 
https://youtu.be/i4JnrAwullI
It could as well be used in some games too to "report damage" or "warn of proximity of edges" etc.

Of course there are always other ways like support SLI setups so one GPU per eye rendering or something else so foveon rendering might not be so critical for performance reasons. 
As I see more of the problem currently the developers who are pushing maybe too much details, or then trying to do fancy graphics (neon lights etc) or simply just not getting model sizes right. 

Anonymous
Not applicable
I think we'll end up with 2 x 2K displays and 120-150 FOV with the CV2 if Oculus don't have foveated rendering nailed and 2 x 4K displays with 12-150 FOV is they manage to get foveated rendering sorted out. I'd be fine with either one tbh as long as they get their depth of field thingummybob working with the CV2. Even with 2 x 2K displays presence will go through the roof with proper depth of field implemented.

I was playing Drop Dead the other day and found myself wishing that I had a headset that offered depth of field for a game like this.