Jon Barron (Google DeepMind)

Barron studied at the University of Toronto from fall 2003 to spring 2007, advised by Sam Roweis, and received an Honours BSc in Computer Science in 2007. As an undergraduate, he interned with the Multisensor Exploitation Team at BAE Systems in the summers of 2005 and 2006, and was a computational biology intern at the Novartis Institutes for Biomedical Research in the summer of 2007.⁠[1][1][2][1][1][1][1]

From fall 2007 to spring 2008, he was a junior research scientist at New York University, hosted by Rob Fergus and Yann LeCun.⁠[1][1]

In fall 2008, he began a PhD in computer science at the University of California, Berkeley, advised by Jitendra Malik, finishing in summer 2013. During his doctorate, he interned at Google Research in the summer of 2008; in 2009 he received a National Science Foundation Graduate Research Fellowship and interned with the Intelligent Robotics Group at NASA Ames Research Center; in spring 2012 he was a visiting PhD student at MIT hosted by Bill Freeman, and that summer he interned at Google[x] under manager Marc Levoy; in 2013 he received the C.V. Ramamoorthy Distinguished Research Award.⁠[1][1][1][1][2][1][1][1][1][1][2]

His doctoral research is represented by SIRFS, which estimates shape, chromatic illumination, reflectance, and shading from a single image of an object; the work was consolidated in a 2015 paper in IEEE Transactions on Pattern Analysis and Machine Intelligence (TPAMI), which subsumes his earlier CVPR 2011, CVPR 2012, and ECCV 2012 papers.⁠[3][3]

In 2013, Barron joined Google[x] as a software engineer, moved to Google Research as a research scientist in 2014, and was promoted to senior research scientist in 2015. At Google he has worked on products including Glass, Lens Blur, HDR+, VR, Portrait Mode, Portrait Light, Maps, and Shopping.⁠[1][1][1][1][1][1][3]

His CVPR 2015 paper “Fast Bilateral-Space Stereo for Synthetic Defocus” embedded a stereo optimization problem in “bilateral space” to quickly solve for an edge-aware depth map used to render depth-of-field effects. In 2016, “The Fast Bilateral Solver,” written with Ben Poole, received a Best Paper Honorable Mention at ECCV 2016; the same year, a burst photography paper he co-authored was published at SIGGRAPH Asia 2016, showing how aligning and merging a burst of images lets mobile phones take good photos in low-light and high-dynamic-range scenes. The technology is used by the Nexus HDR+ feature, and the paper later won a Test of Time Award.⁠[3][3][3][3][3]

With Yun-Ta Tsai he published “Fast Fourier Color Constancy” at CVPR 2017, which learns and evaluates white-balance models in the frequency domain, improving accuracy by 13–20% and speed by 250–3000x; the technology is used by Google Pixel, Google Photos, and Google Maps. In 2018 he became a staff research scientist at Google Research, and that year co-authored the SIGGRAPH 2018 paper “Synthetic Depth-of-Field with a Single-Camera Mobile Phone,” which uses dual-pixel cameras and semantic segmentation to produce shallow depth-of-field effects on phones. In 2019, his single-author paper “A General and Adaptive Robust Loss Function” was a CVPR 2019 best paper finalist; he had two finalist papers at that conference.⁠[3][3][1][1][3][3][3][1]

On March 19, 2020, Ben MildenhallBen MildenhallBen Mildenhall is a computer graphics and 3D computer vision researcher and a co-founder of World Labs.Open the full entry, Pratul P. Srinivasan, Matthew Tancik, Barron, Ravi Ramamoorthi, and Ren Ng submitted the paper “NeRF: Representing Scenes as Neural Radiance Fields for View Synthesis” to arXiv, which trains a tiny non-convolutional neural network with volume rendering to reproduce a scene and synthesize photorealistic novel views. The paper received a Best Paper Honorable Mention at ECCV 2020 and was selected as a Communications of the ACM Research Highlight. The same year, Barron received the PAMI Young Researcher Award.⁠[4][4][3][3][2]

In 2021, Barron became a senior staff research scientist at Google Research, managed by David Salesin. As first author he published mip-NeRF, which showed that NeRF is aliased and anti-aliased it by casting cones and prefiltering the positional encoding; the paper received a Best Paper Honorable Mention at ICCV 2021. In November of that year he submitted mip-NeRF 360, extending mip-NeRF to unbounded real-world scenes, which was presented orally at CVPR 2022; at the same conference, Ref-NeRF, which he co-authored and which explicitly models reflections, received a Best Student Paper Honorable Mention.⁠[1][1][5][3][3][6][3][3][3][3]

In April 2023, as first author, he submitted Zip-NeRF, which combines mip-NeRF 360 with grid-based models such as Instant NGP to reduce error rates by 8–77% and accelerate training by 24x; the paper was a best paper finalist at ICCV 2023. That year he also received an ICLR 2023 Outstanding Paper Award.⁠[7][3][3][1]

In 2024, he moved to Google DeepMind as a senior staff research scientist. That year, papers he co-authored received an Honorable Mention at SIGGRAPH 2024 and an oral presentation at NeurIPS 2024: the former distills a Zip-NeRF into a tiled set of MERFs so that radiance fields can be explored on laptops and smartphones at 60 frames per second; the latter, CAT3D, is a single model built around diffusion and NeRF that handles tasks such as text-to-3D and image-to-3D.⁠[1][1][1][3][3][3][3]

Since 2025, Barron has been a principal research scientist at Google DeepMind. His research interests include computer vision, deep learning, generative AI, and image processing, and most of his research concerns inferring properties of the physical world, such as shape, motion, color, and light, from images, usually with radiance fields.⁠[1][1][3]

At the Runway AIRunwayRunway (Runway AI, Inc.) is a New York–based artificial intelligence company that grew out of the ITP program at New York University’s Tisch School of the Arts in 2018. It develops video generation models and tools for the film, advertising and creative industries, has released the Gen-1 through Gen-4.5 series of models, and has extended its research to “general world models” that simulate the world in real time.Open the full entry Summit in San Francisco, Barron joined NVIDIA Cosmos Lab vice president Ming-Yu LiuMing-Yu Liu (NVIDIA)Ming-Yu Liu is a computer vision and generative AI researcher, vice president of NVIDIA’s Cosmos Lab, and an IEEE Fellow. He leads Cosmos, NVIDIA’s world foundation model platform for “physical AI” such as robots and autonomous vehicles; his team earlier developed image-generation technologies including pix2pixHD and GauGAN (NVIDIA Canvas).Open the full entry and Wayve research director Alexander Toshev on the panel “Grounding Intelligence in the Physical World,” discussing how, as world models have gone from video generators to training environments for agents and robots, the gap between simulation and the real world can be closed.⁠[8][8][8][8]

He has served as an area chair for CVPR, ECCV, and ICCV since 2018, and has been a member of the CVPR 2023 organizing committee and the CVPR 2021 Longuet-Higgins Award committee.⁠[1][1][1]

Personal website: https://jonbarron.info/ (opens in a new window)⁠[2]

X: Public account: https://x.com/jon_barron (opens in a new window)⁠[3]

GitHub: Public account: https://github.com/jonbarron (opens in a new window)⁠[3]

Google DeepMind: Official website: https://deepmind.google/ (opens in a new window).⁠[3]