Note: This video may require joining the NVIDIA Developer Program or login

SIGGRAPH 2019: Deep Learning for Content Creation and Real-Time Rendering- A Style-Based Generator Architecture for Generative Adversarial Networks

Ming-Yu Liu, NVIDIA
We propose an alternative generator architecture for generative adversarial networks, borrowing from stDeep learning continues to gather momentum as a critical tool in content creation for both real-time and offline applications. Join NVIDIA's research team to learn about some of the latest applications of deep learning to the creation of realistic environments and lifelike character behavior. Speakers will discuss deep learning technology and their applications to pipelines for film, games, and simulation.yle transfer literature. The new architecture leads to an automatically learned, unsupervised separation of high-level attributes (e.g., pose and identity when trained on human faces) and stochastic variation in the generated images (e.g., freckles, hair), and it enables intuitive, scale-specific control of the synthesis. The new generator improves the state-of-the-art in terms of traditional distribution quality metrics, leads to demonstrably better interpolation properties, and also better disentangles the latent factors of variation. To quantify interpolation quality and disentanglement, we propose two new, automated methods that are applicable to any generator architecture. Finally, we introduce a new, highly varied and high-quality dataset of human faces

View More SIGGRAPH 2019 Talks and Sessions