The CDRIN at the SIGGRAPH 2023
Synopsis: In this Q&A, Julien Coll, Co-Director, and Vahe Vardanyan, AI Scientific Director, share their experience of SIGGRAPH 2023 in Los Angeles, a significant event in graphics, VFX, and AI.
SIGGRAPH celebrated its 50th anniversary this year, and while the festivities may not have been awe-inspiring, the event did not disappoint us. We were amazed by what we witnessed in terms of academic research, best practices in production, and emerging standards in computer graphics and AI.
CDRIN’s Co-Director, Julien Coll, and AI Scientific Director, Vahe Vardanyan, returned from Los Angeles enriched with new connections, theoretical knowledge, and experiments to further CDRIN’s understanding of the practical application of emerging technologies for SMEs. They also gained insights to help them to stay ahead of future industry standards in the entertainment industry.
Artificial intelligence is experiencing an unprecedented surge, and the technological landscape is evolving at lightning speed. In this context, it is crucial to carefully select the research directions most relevant to the digital creativity sector. This is exactly what we do, drawing insights from academic papers as fostering a profound understanding of studios’ experiences, their needs, and their visions. — Julien Coll, Co-Director
A lot happens at SIGGRAPH, making it impossible to provide a complete account. The Production Sessions, during which industry leaders share their expertise, particularly piqued your interest. Why?
Julien Coll: They offered a concrete insight into how studios strive for a higher excellence level while challenging existing norms. Through these production sessions and interactions with industry players, we furthered our understanding on cutting-edge technologies, their practcality, and their potential applications. SIGGRAPH is a significant event because it provides us with a portion of the raw material from which we construct a clearer vision of the future standards in the entertainment industry.
Vahe Vardanyan: The presentations are delivered by game-changing animation and VFX studios such as Pixar, Disney, and Sony that contribute to blockbuster production. They delve into new technologies, their evolution, and recent applications. The Production Sessions help us steer our research in the right direction and envision concrete, practical, and realistic tools to support the AI transition of Quebec SMEs in the entertainment sector. It’s worth noting that SIGGRAPH, a significant event in the VFX industry, now also emphasis on artificial intelligence.
The CDRIN conducts applied research on various emerging technologies featured at these conferences. Which ones will we explore further after attending SIGGRAPH?
Vahe Vardanyan: We could easily find an academic paper on almost all of the applied research subjects we are conducting. However, our primary interest lies in the advancements presented in 3D, Gen AI, NeRF, synthetic data, and simulation, as we aim to better understand and master them. NVIDIA’s conferences, for example, are highly relevant to generative AI. It is a major player in artificial intelligence applied to the entertainment industry and a serves as a source of inspiration for our future applied research projects. We have also observed advancements in physical and graphical simulation in VFX. For instance, studios are increasingly adopting real-time rendering, which is also one of our expertise areas developed with the researcher-programmer Olivier Therrien. Lastly, studios are combining all these technologies, which holds great promise.
Julien Coll: These cutting-edge technologies are driving a paradigm shift in both the entertainment industry and our fields of expertise [computer vision and rendering].
Generative AI for 3D presents new perspectives for producing highly immersive data with unparalleled realism and intricate detail. While data generation for 2D is already a widely debated topic due to its substantial impact on production, accessibility, and expertise, the potential of generative AI in 3D is equally if not more, promising.
NeRF has generated considerable interest among entrepreneurs for its capabilities in capturing and representing reality. It is an ultra-efficient tool offering a horizon of possibilities, but its application in game design and other fields remains challenging. This technology is so disruptive in production that we don’t yet know how to apply it. How can we improve existing pipelines to integrate it? How can we use it to drive innovation? We aim to stay at the forefront and explore practical applications of NeRF, a technology not yet utilized in production, as it holds the potential to transform our interactions with both virtual and real environments.
Synthetic data is experiencing significant growth in AI projects, and provides noteworthy advantages beyond the entertainment industry.
It contributes, in particular, to reducing the production costs of AI models, as we discuss in our first article published on the Medium platform [The Power of Synthetic Data: Using Games Engines to Improve AI].
What is the impact of this emerging paradigm for CDRIN?
Julien Coll: Besides the fact that our entire team is now deeply immersed in artificial intelligence, the emergence of numerous new technologies and the industry’s rapid evolution have temporarily positioned CDRIN closer to academic research and the experiments conducted by large companies. Currently, our focus is more on exploration and research than on development. This choice, with advantages and disadvantages, is our way of staying at the forefront. It doesn’t alter the essence of CDRIN. On the contrary, our ability to embrace these technologies equips us with the expertise required to support SMEs once the dust settles. That remains our core objective: we are shaping research directions to forge new tools and pipelines that we continually update or streamline based on the latest academic papers and experiments, much like those witnessed at SIGGRAPH. To achieve this, we engage in more frequent and regular exchanges with our partners, bringing us closer to the industry than ever!
Julien and Vahe returned from SIGGRAPH motivated to find that CDRIN’s technological directions are on the right track. Therefore, our exploratory work continues, including a significant public research project — Unlocking Creativity through Digital Intelligence: The Keystone to the Future of Entertainment (we’ll share more details soon!). All the while, we remain committed to offering ways to ethically and responsibly utilize these technologies. It’s not about replacing humans: it’s finding means to enhance quality and reduce the environmental impact of productions.
Recommended Reading
During this international showcase of graphic innovation, we had the privilege of exploring cutting-edge research in AI, 3D rendering, and volumetric modeling. Among some notable discoveries, let’s highlight these three papers recommended by Vahe Vardanyan:
3D Gaussian Splatting for Real-Time Radiance Field Rendering, Inria: This paper explores an innovative method for real-time rendering of radiance fields. By utilizing 3D Gaussian splatting, the authors achieve remarkably high-quality radiometric renderings. This approach opens up new possibilities in the XR sector.
Learning Physically Simulated Tennis Skills from Broadcast Videos, NVIDIA: This impressive study demonstrates the possibility of learning simulated physical skills from broadcast videos, offering intriguing insights for simulation and machine learning.
NeuralVDB: High-resolution Sparse Volume Representation using Hierarchical Neural Networks, arXiv: The authors propose a highly relevant method for representing sparse volumes at high resolution using hierarchical neural networks. This advancement has enormous potential in fields such as medical modeling, realistic simulation, and the creation of high-quality 3D content. It will undoubtedly find applications in our volumetric modeling research.
About the authors
Julien Coll, in his co-director role at CDRIN, is driving a strategic vision focused on R&D and innovation, along the digital creativity corridor along the St. Lawrence River, while fostering closer relationships with Canadian markets. With over 10 years of experience as a 3D technical artist and business development advisor, he plays an essential role within CDRIN by overseeing AI projects related to entertainment and healthcare. His expertise spans audiovisual, 2D/3D, video games, VFX, animation, scripting, management, and events. Involved in influential industry committees (BCTQ, Quebec Video Game Guild, MIGS, CAFÉ, and Siggraph Montreal Chapter), he actively contributes to shaping the future of innovation.
Dr. Vahe Vardanyan, Scientific Director, spearheads CDRIN’s AI team, directing projects spanning computer vision, graphics, and more. With a decade of experience in the digital creativity sector, he’s crafted VR applications, games, and museum exhibits at various Montreal firms. Previously, he was an AI researcher at Ubisoft La Forge and a postdoc at Mila, collaborating with Prof. Yoshua Bengio on climate change visualization via GANs. His present mission is to facilitate AI research for small businesses and academics, uniting these realms and translating R&D into real-world applications.
About CDRIN
The Centre de développement et de recherche en intelligence numérique (CDRIN) conducts research and technology transfer in digital intelligence, particularly in entertainment technology and artificial intelligence for small, medium, and large-sized companies. It also responds to needs in skill development and workforce training in this rapidly evolving sector.
By strengthening the links between university research and Montreal-based SMEs, the CDRIN will become a vehicle which increases capacities of the applied research culture in digital creativity throughout Quebec. The CDRIN will support the implementation of a digital creativity pole in Eastern Quebec to provide an environment that can boost this type of initiative in our region.
CDRIN is a member of the CCTT network and affiliated with Cégep de Matane since 2011, CDRIN has developed recognized expertise in animation, modeling, rendering, artificial intelligence and synthetic data generation.
Contact us at info@cdrin.com | Follow us on LinkedIn
