[x3d-public] AI Video Codec...

[Valentin Christoph]

Hi Kevin,

Yes, if we have a look “from a distance”, then the principle is quite similar to what is done with avatars at the Web3D Consortium.

In the HAnim standard (which the Web3D Consortium is responsible for, too), we have also a possibility to transmit the major part of the data only once, when the avatar is loaded (added to the scene), and afterwards only the state is transmitted (low-bandwidth).

However, the YouTube video you are referring to, is quite different from HAnim, if it gets to the details.

Mainly, I guess those guys cannot provide a 3-dimensional (or even 4-dimensional) model of the virtual reality in question, which is advantageous in many cases.

Happy easter,
Christoph

Von: Wriston Kevin <Kevin.Wriston at kontron.com>
Gesendet: Sonntag, 21. März 2021 22:45
An: Valentin Christoph <Christoph.Valentin at kontron.com>
Cc: Onic Roman <Roman.Onic at kontron.com>; Beicht Peter <Peter.Beicht at kontron.com>
Betreff: AI Video Codec...

Hi Chrstioph,

This reminds me of your efforts with X3D.  Not sure if you would calls this AI... or even a codec... but it's pretty crazy the implications.

I guess that this can only really be applied to a video conferences at the moment given the problems that they have with introduction of new/foreign objects to the stream.  But maybe in the future there is a balance between new i-frames and metadata.  Or maybe the combination of better compression and faster networks wins in the end.  🙂

Saludos,
Kevin

https://youtu.be/dVa1xRaHTA0<https://eur04.safelinks.protection.outlook.com/?url=https%3A%2F%2Fyoutu.be%2FdVa1xRaHTA0&data=04%7C01%7CChristoph.Valentin%40kontron.com%7C58a6a3035c0c405b3d8408d8ecb2a48b%7C8c9d3c973fd941c8a2b1646f3942daf1%7C0%7C0%7C637519599296609963%7CUnknown%7CTWFpbGZsb3d8eyJWIjoiMC4wLjAwMDAiLCJQIjoiV2luMzIiLCJBTiI6Ik1haWwiLCJXVCI6Mn0%3D%7C1000&sdata=ChfLE564x3S9lpvbF%2FehdvLKfjY4orc6bRedVDaEp%2B0%3D&reserved=0>
[https://i.ytimg.com/vi/dVa1xRaHTA0/maxresdefault.jpg]<https://eur04.safelinks.protection.outlook.com/?url=https%3A%2F%2Fyoutu.be%2FdVa1xRaHTA0&data=04%7C01%7CChristoph.Valentin%40kontron.com%7C58a6a3035c0c405b3d8408d8ecb2a48b%7C8c9d3c973fd941c8a2b1646f3942daf1%7C0%7C0%7C637519599296609963%7CUnknown%7CTWFpbGZsb3d8eyJWIjoiMC4wLjAwMDAiLCJQIjoiV2luMzIiLCJBTiI6Ik1haWwiLCJXVCI6Mn0%3D%7C1000&sdata=ChfLE564x3S9lpvbF%2FehdvLKfjY4orc6bRedVDaEp%2B0%3D&reserved=0>
NVIDIA’s AI Puts Video Calls On Steroids! 💪<https://eur04.safelinks.protection.outlook.com/?url=https%3A%2F%2Fyoutu.be%2FdVa1xRaHTA0&data=04%7C01%7CChristoph.Valentin%40kontron.com%7C58a6a3035c0c405b3d8408d8ecb2a48b%7C8c9d3c973fd941c8a2b1646f3942daf1%7C0%7C0%7C637519599296619958%7CUnknown%7CTWFpbGZsb3d8eyJWIjoiMC4wLjAwMDAiLCJQIjoiV2luMzIiLCJBTiI6Ik1haWwiLCJXVCI6Mn0%3D%7C1000&sdata=KoUU6TeyDz17UmelqahpTLxz5lbtT%2FKU6m%2Fe%2BIl8eXU%3D&reserved=0>
❤️ Check out Weights & Biases and sign up for a free demo here: https://www.wandb.com/papers ❤️ Their mentioned post is available here: https://wandb.ai/ayus...
youtu.be


-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://web3d.org/pipermail/x3d-public_web3d.org/attachments/20210405/4200c644/attachment-0001.html>