NVIDIA AI MEETING TECHNOLOGY...
Updated: Oct 8
They say that necessity is the mother of invention and this certainly seems true for Nvidia who may have come up with some game-changing technology for videoconferencing. This could be the future, I'm not exaggerating.
As lockdown has gripped, more and more of us have found ourselves relying on video call technology for both work and home. The issue has always been around the bandwidth demands video calls put on networks and now Nvidia thinks they have found a way of reducing these demands by 'an order of magnitude'.
For smooth, jitter-free video your computer/phone takes a series of still pictures and then sends them over and over again at 24 frames per second to produce a clean video. The problem is, these images are large and there is a lot of them which is why the video sometimes stutters as the network struggles to keep up with the demands placed upon it.
The easiest fix is to increase your network bandwidth, but what if this isn't an option? Nvidia may have come up with the solution. Instead of sending 24 pictures per second, they have come up with technology that lets you send just 1 picture and then the software manipulates it using small points that focus on the eyes, nose and mouth. The bandwidth required to update these points is tiny so the videos are consistently smooth.
Another advantage of the software is that it can automatically adjust the image so participants whose camera is off to the side can be reorientated to face forward. Oh, and it can use your face movements to animate avatars in the same way Apple lets iPhone users with their Memoji's.
Looking at the video, the head movement is very controlled which may hint at the limitations of the current version of the software but it is a great start. Judge for yourself, click on the image below to watch the video.