FAQ

Tier3D is a set of hardware gadgets and an AI software cloud service. Tier3D hardware gadgets have special design and sensors to capture the best possible data for the AI cloud. The AI cloud enables novel AI apps on the desktop, phone, watch and glasses, allowing people to create content and games without any 3D modeling or programming.

Tier3D Phone, watch and glass capture the best possible 3D video and ambient light field. The ambient lightfield capture gives an estimate of the true texture and the material of the objects independent of the lighting. This enables better AI understanding as well as an ability to re-render the scene, including new objects as needed, in a realistic way. Over the time as the system becomes more and more accurate, it would be able to render correct shadows and reflections – for example, a reflection of the virtual object in a real mirror.

The other aspect of the capture is the behavioral modeling. The Tier3D watch can do a continuous capture in the passive mode and understand and model behavior of everything in your surroundings. In an AI system, the larger size of data improves the accuracy of the models so the crowdsourcing of data from millions of devices allows a detailed understanding of the behavior of everything being observed. Over time the AI will become more and more accurate and we envision a time when you will find it hard to distinguish between interaction with a model of your friend vs interaction with your real friend. The goal is to be both photorealistic and behavior realistic.

The accurate models of everything in the world will enable generation of videos from scripts and prediction of next states based on AI. So if you capture your trip to a beach and then add your best friend to it using MindStudio, the software will be able to calculate an alternate timeline based on their presence.

This will also enable real-world gaming and you will be able to convert a capture of the real-world into a game. The system will able to calculate the continuously evolving timeline based on the actions of the real people. If a real person will be unhooked from the game, they could be replaced by their avatar, who will mimic their behavior as closely as possible to the real user.

There are other applications of the time-evolution AI, like in robotic automation. One such application is in autonomous cars, where a car can plan the current action based on the time-evolution of the most probable and worst case scenarios. Such time-evolution is extremely important for autonomous cars because an action based only on the immediate inputs may not be the best action over a longer duration of time. It’s similar to chess in which the capture of the queen in the next move may lead to a checkmate in the next 7 moves.