Eye tracking module and Foveated Rendering


Those are consumer glasses. I’m basing what I’m saying on the higher end research level stuff.


Do you have any link where they show the benefit of 500 hz tracking for VR ?


There isn’t a benefit yet, thats what I’m trying to tell you. lol The Technology is not ready yet. Variable eye sizes, the shape of the socket, the depth level the eye sits at in the socket, all has an impact on the effectiveness of eye trackers even at the highest end.

Even if they get the hardware down, they would have to redesign the entire graphic pipeline to be able fully utilize the benefits of foveated rendering.

“If you can render half of the screen in high quality and the other half in low quality, then that’s going to add up to really important speed gains.”

We already do stuff like this with lens matched shading and the other tools in the VRworks suite. You dont need eye tracking at all to do that.


And again, who says you can’t do it at just the framerate of the panel ?

I’m talking about rendering the area in focus in high quality.


WE ALREADY DO THAT! The center of your view can be rendered in the highest quality, with lower quality in concentric rings throughput the FOV toward the edges.

Eye tracking at a low refresh rate does not work well because your eyes never stop moving. They dont stop moving, and they make micro movements.

If 500hz is not fast enough to capture eye movement quickly, 60hz or 50hz (like commercial eye tracking available on computers today,) lag terribly. They just do.


But that’s not forveated rendering. What if you look at the edges ? That would be bad if they were low quality when the center, where you are not looking, would be high quality.


But why lag ? If they just shoot a pic before the new frame, there is no lag


If a target on a range is moving at a fixed rate of speed, and your hand (with the pistol in it) moves at a much faster, and also erratic rate, (while also micro wobbling and teetering uncontrollably,) the stable speed of the target becomes irrelevant.

Low frame rate eye tracking does not work well because your eyes move too quickly and in erratic fashion for a simple camera to pick it up. The camera is not the cause of the lag, your eyes are the source of the lag


Arent you saying that it doesnt make sense to do it on a display panel under 500 hz ? Cause if your eyes really move that quick, it’s going to be incorrect a lot of times anyway because your eyes move between each frame. At least I think that’s what you’re saying, right ?




That’s from 2010 and not geared at VR


it still gets the basics across.

There is the refresh rate of the camera looking at your eye, the update rate of the tracker collecting info on data points the software paints on your eye, etc.


I still haven’t seen any evidence that 500 hz for VR is needed for eye tracking in order for it to work correctly. If it was, why would that same company put 120 hz eye trackers in their VR glasses, 7 years after that article ?


Eye tracking being used to move a mouse cursor on your PC does not need to be as accurate as an eye tracker would need to be to track the 3 degree spot on your retina where the fovea region is. totally different applications of the technology have totally different needs.


so you’re saying that company put an inferior, not very good working, product in their glasses ? It just makes no sense


It makes perfect sense. Different products are tailored and sold for different market trends and needs. Look at the Pimax headset itself. They have the 4k (low end) 8k (mid range with scaler) and 8k X (top of the line requiring top of the line.)

That’s just how industry works man. It sucks, but thats the game they play.


But you’re claiming that 120 hz is not good enough for VR and that that company is thus selling rubbish. I highly doubt that’s true


120hz is ok for certain things, but not for others.


I sure want an eye tracking as the mouse for my Project Cars 2…that would be cool to use your eyes to choose your selection than using your mouse.


I’ve been thinking: in attempts to overcome the inherent latency involved in retinal tracking, could the foveated area be expanded in attempt to compensate for the current limitations? A larger rendered area wouldn’t conserve as much gpu power, but it could somewhat, while also providing the benefits of more realistic avatars/social interaction, and other features companies like Tobii have been demonstrating recently…

Does it need to be perfect in order for it to be useful for VR?..