In nuce: A research team has presented IDE-3D – a new generative model that can be used to draw, edit and control faces. IDE-3D stands for Interactive Disentangled Editing for High-Resolution 3D-aware Portrait Synthesis: Interactive Disentangled Editing for High-Resolution 3D-aware Portrait Synthesis). The model works on the basis of GAN, the Generative Adversarial Network (see also the article “Canvas/GauGAN: Painting with Artificial Intelligence” from DP 06 : 2022, in which our author Ralf Gliffe explains how the painting programme Nvidia Canvas makes the possibilities of GAN currently usable).
IDE-3D: Interactive Disentangled Editing for High-Resolution 3D-aware Portrait Synthesis
In toto: The summary of the academic paper states that there is a dilemma when generating 3D faces – either they are easy to edit but have a low resolution, or they are difficult to edit but have a high resolution. IDE-3D provides a remedy here. The complete paper can be viewed at mrtornado24.github.io .
Who is involved in the research? A team of six people from the Tencent AI Lab, the ByteDance company and Tsinghua University are involved in the research. Tencent AI Lab is a consortium of more than 100 scientists and over 300 engineers. The internet technology company ByteDance is best known in Germany as the developer of the TikTok platform. Finally, Tsinghua University is one of the most renowned universities in China. Doctoral students Yichun Shi, Lizhen Wang, Jingxiang Sun, Research Manager Jue Wang, Chief Researcher Xuan Wang and Professor Yebin Liu are working on IDE-3D.
Clicked on: As recently as Friday 4 November this year , Digital Production reported on exciting research results – this time from Otto von Guericke University Magdeburg.
Source: 80.lv ( report by Theodore McKenzie)
