Suggestions
Michael J. Black
Max Planck Director, Chief Scientist at Meshcapade, Spokesperson at Cyber Valley, German & Swedish Academies of Science, Scientist, Entrepreneur
Michael J. Black is a prominent figure in the field of artificial intelligence and computer vision, particularly in the area of 3D digital humans. He is the Co-Founder and Chief Scientist of Meshcapade, a startup specializing in AI solutions for modeling, analysis, and synthesis of 3D digital humans in motion.1
Professional Background
Black is a founding director of the Max Planck Institute for Intelligent Systems and an honorary professor at the University of Tübingen.3 He has an extensive academic background, having received his B.Sc. from the University of British Columbia, M.S. from Stanford, and Ph.D. from Yale.2 Throughout his career, he has held positions at various institutions, including the University of Toronto, Xerox PARC, Brown University, and Amazon.2
Entrepreneurial Ventures
In addition to his role at Meshcapade, Black has been involved in other successful entrepreneurial ventures:
- Co-founded Body Labs Inc. in 2013, which was later acquired by Amazon in 2017.12
- Served as a Distinguished Scientist at Amazon from 2017 to 2021.2
Research and Innovations
Black is renowned for his pioneering research on virtual humans, which formed the foundation for Meshcapade.3 He is a co-inventor of the patented SMPL body model technology, which has become the de facto standard representation of 3D humans.1
Accolades and Memberships
His contributions to the field have earned him numerous prestigious awards and recognitions:
- Recipient of the PAMI Distinguished Researcher Award2
- Winner of all three major test-of-time awards in Computer Vision2
- Member of the German National Academy of Sciences Leopoldina2
- Foreign member of the Royal Swedish Academy of Sciences2
Current Roles
As of 2024, Michael J. Black holds several key positions:
- Co-Founder and Chief Scientist at Meshcapade13
- Director at Max Planck Institute for Intelligent Systems26
- Speaker of Cyber Valley, Europe's largest AI ecosystem3
Black's LinkedIn profile (username: michael-j-black-41751114) reflects his current roles and his ongoing involvement in the development of AI and 3D digital human technologies.45
Highlights
How will AI change storytelling in film and beyond? I was part of a great #SIGGRAPH2024 panel exploring this. TechCrunch has summarized it here:
Tick tock... 2:20 today at #SIGGRAPH2024, I'll present "ContourCraft: Learning to Resolve Intersections in Neural Multi-garment Simulations." Mile High 4. Artur Grigorev, the lead author, didn't get a visa so I'm the designated hitter. What's it about?
Neural clothing is coming but current methods for learned garment simulation don't handle cloth-cloth interactions well.
ContourCraft is the first learned method that can successfully simulate neural outfits with an arbitrary number of layers.
ContourCraft is a graph neural network (GNN) that autoregressively predicts nodal accelerations, similar to our previous method, HOOD.
However, we apply two key modifications that enable the model to both prevent and resolve cloth interpenetrations.
A central contribution is a new Intersection Contour (IC) loss term that penalises the lengths of cloth penetrations. With the IC loss, the model learns to resolve intersections between layers of fabric.
ContourCraft is trained to minimize the combined physical losses. After being trained with Intersection Contour loss, our model can resolve intersections (red curves in the video) on the fly. It can even start the simulation from severely intersecting geometries.
Importantly, we can automatically resize and repose our outfits and ContourCraft resolves any penetrations as the simulation goes on.
Neural cloth capture and simulation is the future. We’re preparing code to automatically convert ContourCraft simulations into Alembic format so that they can be easily imported into CGI software.
Authors: Artur Grigorev, Giorgio Becherini, Otmar Hilliges and Bernhard Tomaszewski from Max Planck Institute for Intelligent Systems and ETH Zürich.
arXiv: arxiv.org/abs/2405.09522 Project: https://lnkd.in/gXtFMbSF Video: youtu.be/5i9DXhPMEow
Code will soon be released, follow Artur Grigorev to get notified.
https://lnkd.in/gB4wSM8K
