Another interesting example of this is Meta's Kodak Avatars where currently they require you to go to a lab scan your face but the end result is a very realistic 3D rendering which is better able to bridge the divide between interacting with someone over the internet instead of in person. Here's a vid of Lex Fridman and Zuck showing it off https://www.youtube.com/watch?v=MVYrJJNdrEg
jinweiwong
It's fascinating how VR headsets are able to render your entire face even though a large chunk of it would be occluded by the headset itself. It does seem like this has the potential of making telecommunication more interactive.
weinatalie
Since VR headsets (ostensibly) only cover one region of your face, I wonder how they are able to render your entire head. Is the headset able to see your whole expression, or does it interpolate an expression given several existing scans of your face? Can it pick up your audio and create an accurate lip-sync without having seen your mouth?
llejj
The eye region looks pretty clear, while the rest of the face looks blurred. Is this because the headset has inward facing cameras for the eyes, and just estimates the rest of the face?
OnceLim
After seeing previews of video chatting on Apple Vision Pro, it is clear we got a long ways to go to make this actually viable in real professional conferences. Since it is basically draws your face into the video, there can be some ethical problems that may arise. For example, there are many facial recognition softwares that clearly recognizes only certain racial groups, and the same can go for Apple Vision Pro.
AlsonC
Does the blurryness of the area not surrounding the eye have to do with the vision pro choosing to make up missing compute by using less quality on the areas further from the eye, mimicking the peripheral vision that our eyes employ?
aravmisra
After some digging, I found that this engineer/researcher's team at Meta worked on the face tracking tech for Meta's avatar system- https://www.linkedin.com/posts/sofienbouaziz_real-time-avatar-calls-for-when-youre-not-activity-7086907874513195009-mflD/?trk=public_profile_like_view
The post isn't informative but upon looking at his profile, it seems that it's a Computer Vision based software. Pretty interesting and obviously a booming field with all the applications.
ShonenMind
It's worth noting that the HAIRSTYLES for all these avatars are actually "fake", in that these hairstyles are the exact same from when their faces were scanned by the vision pro. Furthermore, whenever their head nods or shakes, the hair doesn't move freely. Rather, the hair is pretty rigid, and doesn't quite change from it's normal position.
sparky-ed
I think what is really interesting about VR/XR is that it represents a significant leap in how we see and interact with the virtual environment. vision pro's ability to make things photorealistic avatars that mimic physical presence. I wonder how it maintains to keep privacy and data implication while extracting physical presence of the user.
Another interesting example of this is Meta's Kodak Avatars where currently they require you to go to a lab scan your face but the end result is a very realistic 3D rendering which is better able to bridge the divide between interacting with someone over the internet instead of in person. Here's a vid of Lex Fridman and Zuck showing it off https://www.youtube.com/watch?v=MVYrJJNdrEg
It's fascinating how VR headsets are able to render your entire face even though a large chunk of it would be occluded by the headset itself. It does seem like this has the potential of making telecommunication more interactive.
Since VR headsets (ostensibly) only cover one region of your face, I wonder how they are able to render your entire head. Is the headset able to see your whole expression, or does it interpolate an expression given several existing scans of your face? Can it pick up your audio and create an accurate lip-sync without having seen your mouth?
The eye region looks pretty clear, while the rest of the face looks blurred. Is this because the headset has inward facing cameras for the eyes, and just estimates the rest of the face?
After seeing previews of video chatting on Apple Vision Pro, it is clear we got a long ways to go to make this actually viable in real professional conferences. Since it is basically draws your face into the video, there can be some ethical problems that may arise. For example, there are many facial recognition softwares that clearly recognizes only certain racial groups, and the same can go for Apple Vision Pro.
Does the blurryness of the area not surrounding the eye have to do with the vision pro choosing to make up missing compute by using less quality on the areas further from the eye, mimicking the peripheral vision that our eyes employ?
After some digging, I found that this engineer/researcher's team at Meta worked on the face tracking tech for Meta's avatar system- https://www.linkedin.com/posts/sofienbouaziz_real-time-avatar-calls-for-when-youre-not-activity-7086907874513195009-mflD/?trk=public_profile_like_view
The post isn't informative but upon looking at his profile, it seems that it's a Computer Vision based software. Pretty interesting and obviously a booming field with all the applications.
It's worth noting that the HAIRSTYLES for all these avatars are actually "fake", in that these hairstyles are the exact same from when their faces were scanned by the vision pro. Furthermore, whenever their head nods or shakes, the hair doesn't move freely. Rather, the hair is pretty rigid, and doesn't quite change from it's normal position.
I think what is really interesting about VR/XR is that it represents a significant leap in how we see and interact with the virtual environment. vision pro's ability to make things photorealistic avatars that mimic physical presence. I wonder how it maintains to keep privacy and data implication while extracting physical presence of the user.