Deep Generative Mechanisms for 3D: From De Novo Flow Models to Latent Space Optimization

Speaker: Prof. Iro Armeni, Stanford University

Abstract: This talk explores three distinct technical paradigms for generative vision models in 3D reconstruction and synthesis. First, we present a novel 3D rectified flow matching model trained from scratch for robotic assembly, demonstrating how flow-based trajectories can be optimized for precise geometric reasoning. Second, we discuss the architectural adaptation of video diffusion models to enhance 3D Gaussian Splatting (3DGS); by integrating specialized encoding modules into a foundation model that leverage 3DGS priors, we bridge the gap between 2D temporality and 3D spatial consistency. Finally, we introduce a test-time optimization technique for 3D style transfer that utilizes pretrained large 3D generative models to align disparate geometries. Together, these works illustrate a versatile toolkit for modern 3D vision—from designing specialized generative flows to the sophisticated manipulation of large-scale latent priors.

Biography: Iro Armeni is an Assistant Professor at Stanford University, where she leads the Gradient Spaces group, researching at the intersection of Computer Vision, Generative AI, and the built environment. Her work focuses on developing machine perception architectures capable of designing and reconstructing adaptive physical and digital spaces. Previously, she was a Postdoctoral Fellow at ETH Zurich (2020–2023) with a dual appointment in Computer Science and Civil Engineering. Iro earned her PhD from Stanford University in Civil & Environmental Engineering with a PhD Minor in Computer Science. Her multidisciplinary technical foundation includes an MSc in Computer Science and an MEng in Architecture and Digital Design. By bridging generative vision models with architectural engineering, her research group aims to automate the lifecycle of sustainable, data-driven environments. Beyond academia, her work is informed by professional experience as an architect and consultant in both the private and public sectors. Among others, she is a recipient of the Google Research Scholar Program, the ETH Zurich Postdoctoral Fellowship, and the Google PhD Fellowship.

Host: Prof. Francis Engelmann