Computer Graphics World

July-Aug-Sep 2023

Issue link:

Contents of this Issue


Page 23 of 33

22 cgw j u ly • a u g u s t • s e p t e m b e r 2 0 2 3 A rtificial intelligence has the potential to transform every aspect of a post production pipeline. Engineered to max- imize efficiency and foster creativity, AI features are now available across workflows for broadcast, virtual produc- tion, video editing, visual effects, and beyond. Here, we detail innova- tive new AI offerings from three major companies. NVIDIA: New AI techniques for visual effects & animation At SIGGRAPH 2023, NVIDIA released around 20 research papers focused on the advancement of generative AI and neural graphics. Developed in collaboration with over a dozen universities across the U.S., Europe, and Israel, the papers explore new applications for text- to-image technology, inverse rendering tools for transforming still images into 3D objects, neural rendering models for generating re- al-time visual details, and more. Visual effects artists and digital artists may especially benefit from NVIDIA's research into new AI techniques that will automat- ically convert 2D images and videos into complete 3D represen- tations. This will greatly accelerate the process of transforming concept art into fully rendered environments populated with 3D characters and objects. The technology also streamlines the pro- cess of producing 3D characters capable of performing realistic movements and diverse skills—without requiring expensive mo- tion-capture data. When animators generate 3D characters, the process of layer- ing in realistic details can be a very time-consuming and compu- tationally expensive endeavor. Hair is a particularly complex detail to generate. For example, humans have an average of 100,000 hairs on their heads, and each one reacts dynamically to motion and to the environment surrounding it. Animators typically use physics formulas to calculate hair movement, and must oen simplify it accordingly if resources are limited. NVIDIA's new research showcases an AI technique that is able to simulate tens of thousands of high-resolution hairs in real time using neural physics. The process also teaches a neural net- work to predict the movements of each individual hair. Specifically optimized for modern GPUs rather than CPU-based AI: NEW APPLICATIONS FOR POST PRODUCTION INNOVATIVE FEATURES FOR VFX, ANIMATION, VIDEO EDITING, AUDIO EDITING, AND BEYOND. BY KENDRA RUCZAK

Articles in this issue

Archives of this issue

view archives of Computer Graphics World - July-Aug-Sep 2023