Keynote  Speakers

 >> Empowering Artists through Machine Learning: Balancing Creativity and Control 

​

Abstract: Recent advances in generative modelling have enabled the creation of videos, paintings, and songs from simple inputs like text. Models such as Stable Diffusion and SORA have made it possible for anyone to produce impressive results. However, these tools often come at the expense of losing artistic control and creativity.

 

In this talk, we will explore the intersection of technology and creativity. how these models can undermine artists' creative processes and why they might not be ideal for production settings. We will discuss strategies for developing Machine Learning tools that truly empower artists and enhance, rather than hinder, artistic expression.

 

We will also showcase some of the Machine Learning tools developed at Cinesite. These tools have enabled artists to produce higher quality work more efficiently while maintaining full control over their creative processes. Additionally, we will examine common challenges in translating cutting-edge research into practical, artist-friendly tools for production.

.

 

Bio: Riley Miladi is the lead ML researcher at Cinesite, researching AI driven solutions for films and feature animations. With a background in Visual Effects from the University of Hertfordshire, they created physical simulations and procedural setups for film and games. Later they transitioned to a ML research position at Embark Studios focusing on Deep Reinforcement Learning and physically based animation for games. More recently during their position as a researcher at DNEG, they had the opportunity to do further research on Generative Models for animation and crowds for VFX and animation. They currently focus on leading various machine learning projects as well as leading the development of global machine learning strategy at Cinesite.

Riley specialize in deep RL and generative models with an emphasis on ML based animation and human motion but they have contributed to diverse projects in domains such as: geometric deep learning, computer vision, NLP and representation learning. Driven by a deep passion for machine learning, neuroscience, robotics, physics, and art, Riley is eager to research cutting edge solutions to create tools that empower artists and unleash their creativity.

​

​