KIVI: AI-BASED GENERATION OF INTERACTIVE VOLUMETRIC ASSETS

KIVI: AI-BASED GENERATION OF INTERACTIVE VOLUMETRIC ASSETS

AI – Volumetric Research&Development, 2020 – 2023

KIVI (KI-basierte Generierung interaktiver Volumetrischer Assets) is a collaborative research project which deals with the 3D capture of real people, 3D reconstruction in the form of volumetric video and targeted modification. The main goal is the development of efficient AI-based methods for the creation of modifiable and partly interactive volumetric characters.

The demand for computer-generated scenes and content has risen sharply in many areas of application, such as gaming, media, film and TV. In particular, the availability of AR/VR applications in the consumer sector is creating further demand. Volumetric video offers an efficient way of presenting real people authentically and realistically in three dimensions. However, the creation of classic volumetric video is still very time-consuming and the subsequent animation of volumetric assets is also very time-consuming and has so far only been possible to a limited extent.

We want to enhance possibilities and in addition, the authenticity of characters in interactive environments is to be increased by integrating their reaction to the player and the environment. To this end, suitable tools are to be created that significantly simplify the previously time-consuming and costly. process.

The project consortium consists of  the Fraunhofer HHI department „Vision & Imaging Technologies“ and INVR.SPACE GmbH.

KIVI is funded by Investitionsbank Berlin in the ProFIT funding scheme with financial support of European Regional Development Fund (EFRE) and the government of Berlin.