In this keynote, Anton will talk about the evolution of real-time graphics research throughout the last 15 years, focusing on the last 5 years. How path tracing is becoming our everyday life not just in film, but in games too. How the recent advances in compute power made machine learning a revolutionizing computational tool across industries, graphics included. How miniaturization of displays, democratization of content creation, and even the pandemic accelerate the demand for more immersive visual experiences and ubiquitous visual compute. In this golden age of graphics, the keynote will conclude with opportunities and challenging research problems, as well as Intel’s open approach to solving these problems.
Anton Kaplanyan leads graphics research for new discrete GPUs at Intel. His current research interests are in physically based path tracing, neural graphics, scalable rendering, distributed rendering, and differentiable rendering. Prior to joining Intel, Anton worked on perceptual and neural methods for AR/VR at Meta Reality Labs Research, on RTX ray tracing hardware and real-time denoising at Nvidia Research. Anton's PhD in physically based light transport from Karlsruhe Institute of Technology led to a startup in physically consistent light editing for artistic control used at Pixar. Prior to his PhD, Anton had been a game developer at Crytek, working on real-time rendering for CryENGINE 3 and Crysis 2, including Light Propagation Volumes. Anton's recent publications are on neural rendering, advanced light transport, image reconstruction, and shading methods in both real-time and offline graphics.
What Art can tell us about the Brain
Margaret S. Livingstone, Harvard Medical School
Live presentation only
Artists have been doing experiments on vision longer than neurobiologists. Some major works of art have provided insights as to how we see; some of these insights are so fundamental that they can be understood in terms of the underlying neurobiology. For example, artists have long realized that color and luminance can play independent roles in visual perception. Picasso said, "Colors are only symbols. Reality is to be found in luminance alone." This observation has a parallel in the functional subdivision of our visual systems, where color and luminance are processed by the evolutionarily newer, primate-specific What system, and the older, colorblind, Where (or How) system. Many techniques developed over the centuries by artists can be understood in terms of the parallel organization of our visual systems. I will explore how the segregation of color and luminance processing are the basis for why some Impressionist paintings seem to shimmer, why some op art paintings seem to move, some principles of Matisse's use of color, and how the Impressionists painted "air". Central and peripheral vision are distinct, and I will show how the differences in resolution across our visual field make the Mona Lisa's smile elusive, and produce a dynamic illusion in Pointillist paintings, Chuck Close paintings, and photomosaics. I will explore how artists have figured out important features about how our brains extract relevant information about faces and objects, and I will discuss why learning disabilities may be associated with artistic talent. Lastly I will mention some recent studies using neural networks to reveal super stimuli for real neurons.
Margaret Livingstone is Professor of Neurobiology at Harvard Medical School. She has done research on hormones and behavior, learning, dyslexia, and vision. Livingstone has explored the ways in which vision science can understand and inform the world of visual art. She has written a popular lay book, Vision and Art, which has brought her acclaim in the art world as a scientist who can communicate with artists and art historians, with mutual benefit.
Roblox is a global platform for 3D coexperience. Over 17 years it grew to 55 million daily active users across console, desktop, mobile, and VR in 180 countries. In this talk I present some of the unique technical challenges for this system's present and future. These arise due to the combination of constraints from real-time 3D performance, social global social media scale, and 100% user-generated content. For some of these challenges I will describe the solutions created and deployed at Roblox. The others are important new areas of investigation I pose to our Research division and the graphics community.
Morgan McGuire is the Chief Scientist at Roblox, accelerating innovation across the company to build a creative, safe, civil, and scalable Metaverse. Roblox combines social interaction with a dynamic 3D environment and economy. Morgan is also known for work on video games, cloud graphics, esports, ray tracing, and augmented and virtual reality. This spans NVIDIA GPUs, the Skylanders® and Titan Quest® video game series; the Unity game engine; the E Ink display used in the Amazon Kindle®; “the bible” of 3D, Computer Graphics: Principles & Practice 3rd Edition; The Graphics Codex; and the Markdeep document system. Morgan holds faculty positions at the University of Waterloo and McGill University and was a full professor at Williams College. Morgan received Ph.D. and M.S. degrees from Brown University and M.Eng. and B.S. degrees from MIT.