Keynote
Dr Gurprit Singh:

Dr Gurprit Singh:
Dr Gurprit Singh is a Principal Member of Technical Staff at AMD, where he works at the intersection of rendering, optimization, and generative AI. Prior to joining AMD, he was a Senior Researcher at the Max Planck Institute for Informatics in Germany, where he led a research group in visual computing. He earned his Ph.D. from the University of Lyon 1, France, and completed a postdoctoral fellowship from an Ivy League college (Dartmouth College) in the USA. He holds a bachelor’s degree from the Indian Institute of Technology Delhi. Gurprit’s research focuses on bridging the gaps between physically based rendering, probabilistic inference, and generative models, with a particular emphasis on Markov Chain Monte Carlo methods as a unifying framework.
Abstract:
Noise has long been a fundamental building block in computer graphics, shaping textures, patterns, and realism across decades. In this keynote, we trace the evolution of noise — spanning from the art of stippling to physically based rendering. We then transition to the transformative role of noise in modern generative AI, where (correlated) noise scheduling has become a silent yet powerful lever in shaping image generation. We will explore how adjusting noise schedules can dramatically alter generated outputs without retraining or fine-tuning, revealing surprising artistic and technical “control” over deep generative models. This talk will take you to a journey through the visual and mathematical evolution of noise — the invisible hand behind the modern generative AI tools.
Professor Min Chen

Professor Min Chen
Min Chen is a professor of scientific visualization at the University of Oxford and a fellow of Pembroke College. He built his academic career in Wales, particularly at Swansea University, where he rose from research officer to full professor. His research spans data visualization, computer graphics, and interdisciplinary applications, with over 200 publications. Chen has held key editorial and leadership roles in major visualization conferences and journals, and played a significant role in pandemic modeling through the RAMPVIS project. He is a fellow of several prestigious professional societies.
Abstract:
This paper reflects on the need for a solid theory in the field of data visualization. Although many people see visualization as too practical for theory, the authors argue that building theoretical foundations is essential—just like in any scientific field. The authors highlight that theory in visualization can benefit from ideas in other areas like math, psychology, and computer science. While progress in this area has been slow, past breakthroughs in science show that theories often develop over time through many small steps. The paper encourages every visualization researcher, regardless of focus, to contribute to this effort, and it shares insights from key discussions at major conferences on how to move forward.