Hansong Zhang's Resume

Hansong Zhang
hansong@wearality.com
hansong@gmail.com
(408) 390-6723

Areas of Expertise

Software, architecture, and algorithms for visual computing including computer graphics, computer vision, web-based technologies and scalable multicore computing

Experience

5/2015-Present VP of Engineering and Co-founder at Wearality Corporation. Wearality Corporation is a leader in delivering instant, wide-field-of-view, clear, low-distortion, and strain-free mobile VR and AR experiences. Its optics technology is exclusively licensed from Lockheed Martin which represents many years of development according to defense and aerospace application requirements. We aim to bring VR experiences to every smartphone user at very high quality and very low cost, experiences that you can carry with you anywhere you go and share instantly with your friends nearby.

2008-5/2015 Founder & CEO at ScalableVision Inc. Since its inception, ScalableVision Inc. has served as an external research and fast prototyping lab for many of the largest technology companies in the U.S., with excellent reputations for delivering on time and exceeding client expectations.
Select Clients and Projects:

• Intel Perceptual Computing Lab’s hand tracking system (The Verge’s coverage--skip to ~56 seconds): from their very beginning ScalableVision has been a partner to Intel's PerC lab in developing an in-house hand tracking system. The system features a 26 degrees-of-freedom hand model driven in real-time by a human hand, via a depth-capable camera such as Kinect or Intel's own GestureCam. We contributed both algorithms and implementations, while providing a context for positioning the technology based on our broad background in computer vision. Related to this word, ScalableVision also independently produced a platform for testing Sculptus, ScalableVision’s prototype demonstrating free-form creative modeling with hands and depth sensors.
• Microsoft Browser-Based PhotoSynth front-end: ScalableVision implemented the entire front-end viewer for Microsoft's photosynth.net panoramas, using only web-based HTML5/CSS/WebGL technology. We delivered diverse rendering paths for the imagery, including pure CSS, WebGL, and Flash. On the Windows 8 Launch Day, our work went live for the first time when a panoramic background was presented atbing.com that can be panned and zoomed. The code we delivered, due to its well conceived architecture and battle-proven reliability, also found use in other Microsoft products that require similar rendering capabilities.
• AMD WebCL for Chromium: ScalableVision is a partner of AMD in bringing the latest WebCL API to the browser, enabling drastically faster computing in Javascript. We also provide Chromium expertise to AMD in various web-related hardware acceleration effort that makes use of the client’s unique APU architecture, such as their optimization of the WebM codec.
• Google Google Earth: from 2009 to 2012, ScalableVision was responsible for Google Earth's transition to modern mobile platforms. We are the initiator of the application’s mobile-first strategy for graphics based on a mobile- (OpenGL ES2-) centric graphics layer that uses ES2 emulation on the desktop. Related to this work, we contributed to Audi A8's navigation system which was based on Google Earth technology.
• Aechelon Technologies: Aechelon is a leading defense contractor that supplies Image Generators (IGs) the power the flight simulations for U.S. Navy and Air Force. From 2008 to 2009, ScalableVision helped Aechelon optimize and re-architect their graphics engine.
• RightWare: Porting of RightWear's entire C/C++ graphics engine to the browser: RightWare Kanzi is a Unity-like 3D and GUI toolkit. We ported the C- and OpenGL-based Kanzi engine and demo programs to web browsers (JavaScript and WebGL) using Mozilla's (then) bleeding-edge Emscripten toolchain. The client was thrilled by our speed of work and in-browser performance of the port. We also helped the client in business development and communications with Mozilla.
• ScreamPoint Hybrid Native and Browser-based 3D Application for City Management: for the application, we proposed and collaboratively developed a novel hybrid architecture where 3D rendering is done natively, but the GUI is constructed in standard HTML5 and rendered by an embedded browser as an overlay on top. The architecture also enabled viewing file, image and video formats using existing browser plugins, which proved valuable for the application.

6/2007-9/2009 Vice President of Technology at Roblox Corporation. Roblox is a online multiplayer environment for user-created games (UCG), differentiated by its easy-to-use game creation software (Lego-style modeling, behavioral scripting, real-time physics) and a virtual economy. According to comScore, ROBLOX is the #1 ranked entertainment website for kids in the US. We’re currently ranked #1 in Total Visits, Total Page Views, Total Minutes, and Average Visits per Visitor in the kids category.

2003-2007 Senior Technical Lead at Silicon Graphics, Inc., a maker of scalable parallel computers. Reported to CTO of the company and led the Future Technologies Group, an inter-disciplinary effort for combining parallel computing, graphics, and digital media. Work in this period includes:

• Computer graphics (ray-tracing) on massively parallel computers (presentation).
• Co-founded the Manta open-source parallel ray-tracer with University of Utah, which scales to hundreds of CPUs.
• Rendering optimization for multicore CPUs (under Intel grants)
• Parallel system architecture issues and solutions in the context of large (Tera-scale) datasets
• Presented first-in-the-world interactive walkthrough of the famous Boeing 777 dataset (350 million triangles) in December 2004
• Load-balancing and culling architecture for parallel rendering on Multipipe systems (a precursor to SLI/CrossFire today) through transparent OpenGL stream capture, analysis, and optimization.
• Integrated real-time, on-demand block-based wavelet decompression and caching to support visualization of Terabyte data in a ray tracer (with MIT/Whitehead BioImaging Institute)
• Investigated video codecs and multi-processor vision/pattern recognition techniques applied to video, as part of performance characterization of digital media workloads on shared-memory multiprocessors.
• Filed 4 patents

2002-2003 Contributed to nVidia's Cg, world's first language for real-time cinematic special effects.

1999-2002 Graphics Architect at Intrinsic Graphics, Inc., a producer of cross-platform game software for Xbox, PS2, GameCube and PCs.

• Responsible for the design and implementation of Intrinsic's Alchemy Scene Graph, the core technology for the company. The extensible architecture adapts to new hardware devices and has avid followers to this day (2015).
  
  
• Through acquisition, the technology went on to thrive at Vicarious Vision and Activision. It has been used by several dozens of published games. It is also the graphics engine under Google Earth (formerly Keyhole) which 2011, passed the one-billion download mark. Notably, the technology became the backbone architecture for next-gen flight simulators currently deployed by the U.S. Navy and Marines.
  
  
• Two patents.

Education http://www.cs.unc.edu/~zhangh/

1995-1998 Ph.D., Department of Computer Science, UNC-Chapel Hill. Dissertation and publications addressed fundamental problems in computer graphics (visibility determination and image-based rendering). Algorithms invented in thesis was later commercialized by Hybrid Graphics, a Finnish company (acquired by nVidia in 2006). Supported by Silicon Graphics Fellowship, August 1997 to ­July 1998 and Board of Governers Fellowship, University of North Carolina at Chapel Hill, 1997

1992-1995 M.S., Institute of Computing Technology, Chinese Academy of Sciences. Research and thesis centered on distributed volume rendering for medical and industrial applications. See link above for publications.

Publications

[0] Zhang, Hansong. A derivation of image­based rendering for conventional three­dimensional graphics. Journal of Graphics Tools, 4(2):27­36, 1999

[1] Zhang, Hansong, Effective Occlusion Culling for the Interactive Display of Arbitrary Models Ph. D. dissertation, Department of Computer Science, University of North Carolina at Chapel Hill



[2] Zhang, Hansong, Forward Shadow Mapping, Rendering Techniques '98, Proceedings of the 9th EuroGraphics Rendering Workshop, Vienna, Austria, June 29­July 1, 1998, Drettakis, G., Max, N. (eds.) ISBN 3­211­83213­0

[3] Zhang, Hansong, D. Manocha, T. Hudson, and K. Hoff, Visibility Culling Using Hierarchical Occlusion Map, Proceedings of SIGGRAPH'97, Los Angeles, August 1997

[4] Zhangh, Hansong, and K. Hoff, Fast Backface Culling Using Normal Mask, 1997 Symposium on Interactive 3D Graphics, ACM SIGGRPH

[5] Bastos, Rui, M. Goslin and H. Zhang, Efficient Radiosity Rendering using Textures and Bicubic Reconstruction,1997 Symposium on Interactive 3D Graphics, ACM SIGGRAPH

[6] Kumar, Subodh, D. Manocha, H. Zhang and K. Hoff, Accelerated Walkthrough of Large Spline Models, 1997 Symposium on Interactive 3D Graphics, ACM SIGGRAPH

[7] Zhang, Hansong, Pattern Generation with Color Map Gouraud Shading, Computer & Graphics, No.1, Vol.20, Elsevier Science Ltd, Jan. 1996

[8] Zhang, Hansong, Order of Pixel Traversal and Parallel Volume Ray­tracing on the Distributed Shared Volume Buffer, Eurographics Workshop on Visualization 1995, Springer Verlag.

[9] Zhang, Hansong, A Traditionalist View of 3­D Image Warping, TR97­043, Department of Compute Science, UNC­Chapel Hill, Nov. 1997

[10] Zhang, Hansong, Volume Rendering for Data on Irregular Meshes, M.S. Thesis (in Chinese), Institute of Computing Technology, Chinese Academy of Sciences, P. R. China