Building performing AR products · Meta/Snap/TikTok AR Ambassador · Clients: Tiffany, Elie Saab, Burberry, Forbes, JLo · https://t.co/rjt1c9z5VC
Feb 15 • 9 tweets • 4 min read
OpenAI just introduced SORA: new text2video #AI model.
While everyone is just posting mind-blowing examples, I want to talk about the consequences.
A thread 🧵🔽
1. Clearly, this isn't the end of the road—generation quality and consistency (the ability to maintain story, style, and details) will get better, and costs will go down.
Videos today are already hard to distinguish from reality, and soon it'll be outright impossible, even with algorithms or other neural networks.
Jan 2, 2023 • 5 tweets • 2 min read
1/4 So what's the idea of the "universal #AR tracker" I'm working on?
We have:
- Surface tracking (including multiple points on vertical and horizontal surfaces)
- Image tracking
- Spatial (device) tracking
- Head / Face / Hand / Body / etc.
2/4 I'm working on a wrapper that allows seamlessly passing objects from one to another, let's say grab an object by hand from surface/image/space in #AR.
To do it, all objects are processed in world space and the wrapper has methods to smoothly attach them to various trackers.
