Changchang Wu
Also known as C. Wu
Changchang Wu is a computer vision researcher whose work connects structure-from-motion software with generative 3D product views. He completed a 2010 Ph.D. focused on geometry-driven feature detection and helped develop multicore bundle adjustment for large 3D reconstructions. His single-author 2013 paper pursued linear-time structure from motion; in 2018, VisualSFM earned the PAMI Mark Everingham Prize, and he authored a Google guide to VR180. As of October 2026, he is affiliated with Google LLC in AI and image processing; in 2025, he contributed to generative AI for shoppable 3D product views from a few images.
Timeline
-
2010
A computer vision doctorate
He completes a Ph.D. focused on geometry-driven feature detection in 2010.
-
2011
Multicore bundle adjustment
He and Sameer Agarwal, Brian Curless, and Steven M. Seitz publish Multicore Bundle Adjustment at CVPR in 2011. The system uses multicore CPUs and GPUs to solve large-scale 3D scene reconstruction problems.
-
2013
Linear-time structure from motion
He presents a single-author paper, Towards Linear-time Incremental Structure from Motion, at 3DV in 2013; the conference program lists his affiliation as Google.
-
2018
The PAMI Mark Everingham Prize
He receives the PAMI Mark Everingham Prize for VisualSFM in 2018; its citation recognizes the software library's contribution to Structure from Motion.
-
2018
Engineering for VR180
He authors Google's VR180 publishing guide as a software engineer in 2018. It explains the format's video metadata standard.
-
May 2025
Generative 3D product views
In May 2025, he contributes to Google's generative AI techniques for shoppable 3D product visualizations from as few as three product images.
-
As of October 2026
AI and image processing at Google
As of October 2026, he is affiliated with Google LLC in AI and image processing.
Career
- Since 2012GoogleSoftware engineer
- 2011–2013University of Washington
- 2008–2011University of North Carolina at Chapel Hill
Most-cited works
- Towards Linear-Time Incremental Structure from Motion
- Multicore bundle adjustmentWith Steven Seitz
- Building Rome on a Cloudless DayWith Marc Pollefeys and Svetlana Lazebnik
- Modeling and Recognition of Landmark Image Collections Using Iconic Scene GraphsWith Svetlana Lazebnik
- 3D model matching with Viewpoint-Invariant Patches (VIP)With Marc Pollefeys
Sources and credits
Photo: We have no published photo of Changchang Wu yet. Send us a free-licensed photo.





