Developer?

hoppingbunny123
Posts: **391** Nexus 6

in General

how people gaze at the horizontal cell phone in front of them is applicable to showing how to use eye tracking for foveated rendering.

here is the picture then i will talk some more;

the two circles at the bottom are the eyes in front of the horizontal phone.

the large horizontal rectangle is the phone display.

the medium vertical rectangle is the part of the display the eyes move around.

the small rectangle in the vertical rectangle is the area we watch.

if the display moves we keep our eyes on the area of the small rectangle, this is in focus.

if the eyes look above the display and the head is still at the spot it was when we were focused on the phone what we look at in the distance above the phone isnt in focus.

if we look away from the phone we create the three rectangles again and focus our eyes on the area of the small rectangle.

this means we look where we situate our head. if we see the exact area of the small rectangle as we hold the phone horizontally as we watch a video, then the small rectangle is the correlation to the head position and the spot we are focused on with our eyes.

then we move our heads again and this same small rectangle is the focus of our attention and eyes, and if we keep that head position and look away from the small rectangle this isnt in precise focus.

this means if you find the small rectangle spot in relation to the head, you can see when the head changes enough it indicates a new small rectangle position, theres a threshold of head movement that indicates the small rectangle position has changed. then you plot the head position and use foveated rendering for the area of the medium and small rectangle.

here is the picture then i will talk some more;

the two circles at the bottom are the eyes in front of the horizontal phone.

the large horizontal rectangle is the phone display.

the medium vertical rectangle is the part of the display the eyes move around.

the small rectangle in the vertical rectangle is the area we watch.

if the display moves we keep our eyes on the area of the small rectangle, this is in focus.

if the eyes look above the display and the head is still at the spot it was when we were focused on the phone what we look at in the distance above the phone isnt in focus.

if we look away from the phone we create the three rectangles again and focus our eyes on the area of the small rectangle.

this means we look where we situate our head. if we see the exact area of the small rectangle as we hold the phone horizontally as we watch a video, then the small rectangle is the correlation to the head position and the spot we are focused on with our eyes.

then we move our heads again and this same small rectangle is the focus of our attention and eyes, and if we keep that head position and look away from the small rectangle this isnt in precise focus.

this means if you find the small rectangle spot in relation to the head, you can see when the head changes enough it indicates a new small rectangle position, theres a threshold of head movement that indicates the small rectangle position has changed. then you plot the head position and use foveated rendering for the area of the medium and small rectangle.

0

It looks like you're new here. If you want to get involved, click one of these buttons!

## Comments

391Nexus 6the algorithm for how people move their eyes focus when watching a video on their phone is;

- the small rectangle

- below the small rectangle in the vertical rectangle

- to the side of the vertical rectangle in the large horizontal rectangle.

then you see them move their eyes back to the small horizontal rectangle, you can predict with some high percentage accuracy they will look at the bottom area of the vertical rectangle next.

add to that a jab of the arm darting out like a boxer and that is the large horizontal position, off to the far side, this is followed by going to the small rectangle area inside the vertical rectangle. replace the arm jab for typing then looking at the screen after typing its the same thing.

it loops over and over again. I bet people have a fingerprint type of structure that's unique to them for how they loop it.

so this algorithm will help with the foveated rendering. you can predict where to send to focused pixels next.

1,550Wintermute391Nexus 6- small rectangle = norm

- vertical rectangle = crude norm

- large rectangle = not norm

There's these three types of norm going on in every video that tells a story. If so there's an algorithm to going through the three norms, and the algorithm i described is at least 1 of the algorithms.

Is the action the norm? Then its in the small rectangle.

391Nexus 6If the thing being looked at is considered as being still by the looker, the norm and crude norm are looped.

Which could be considered algorithm 2.

- Algorithm 1 = thing seen is considered as moving, norm, crude norm, not norm; looped

- algorithm 2 = thing seen is considered as not moving, norm, crude norm; looped

Then algorithm 2 stops and you go back to algorithm 1 .

391Nexus 6so it stood to reason there's a horizontal bias too, and in some videos like the b13 trailer there is a horizontal bias with lots of constant action scenes.

so I considered the back and forth between either side of the vertical rectangle on the large horizontal rectangle the action algorithm, its algorithm 3.

here's how the three algorithms fit together into algorithm 4;

- Algorithm 1 = thing seen is considered as moving, but not constantly moving, norm, crude norm, not norm; looped

- algorithm 2 = thing seen is considered as not moving, norm, crude norm; looped

- Algorithm 3 = thing seen is considered as moving, constantly moving, not norm; looped

- algorithm 4 = algorithm 3, algorithm 2, algorithm 1

algorithm 5 is the social norm is to have two or more people in a group share algorithm 4 in sync. this shows familiarity and acceptance, which would be one algorithm 4.

contrary to this is two or people in a group out of sync in algorithm 4, which would be two algorithm 4.

if the person is apprehensive, they are in a strange group not a family that shows the norm is being out of a family and so out of sync and that brings in 2 algorithm 4. so the family or what is wholesome there's one algorithm 4 in a group, but for unwholesome people the group is multiple algorithm 4 out of sync with other peoples algorithm 4.

so the logic is first if there is algorithm 5 wholesome or unwholesome, that dictates how multiple algorithm 4 will work together, which shows if in sync the algorithms 1, 2, 3, going through the three types of norm.

391Nexus 6the algorithm 4 goes from, algorithm 3 to algorithm 2, for both wholesome and unwholesome in algorithm 5.

the difference comes from where algorithm 2 goes, that changes wholesome to unwholesome in algorithm 5.

- algorithm 2 should go to the small rectangle in algorithm 1, that's familiar, expected, wholesome.

- else algorithm goes to algorithm 3, that's unexpected, unfamiliar, unwholesome. like lucy pulling the football when Charlie brown goes to kick it.

after watching music videos they only have the wholesome algorithm 5 towards the end of the video, the rest of the video is unwholesome, like korn's video freak on a leash.

so for music videos I would make another algorithm, algorithm 6.

algorithm 6 which is for media like korns music videos, has much more unwholesome algorithm 5 during the video followed by the wholesome algorithm 5 ending the video. you can see time and time again the trust and the antithesis of this trust, like lucy holding the football Charlie browns going to try to kick, in the bulk of the video. so if you have the foveated rendering and see the time of the video isn't near the end, algorithm 2 will probably come from then go to algorithm 3 again before going from algorithm 2 going to algorithm 1.