Kun Zhou: "Ants" challenging "Superman", from legend to reality

RESEARCH

May 16, 2011|admin |Source: 研究生院英文网

figure1. Kun Zhou(the first one from right side) and his colleagues

figure2.a image rendered by Render Ants

Who will drive the next round of technological revolution in the coming decades, and show us the beauty of “technology is changing the life”? The annual selection of “Outstanding Young Innovators of the world”, by “Technology Review” of MIT, is a weathervane. Just released this year, Professor Kun Zhou, the 33-year-old Scholar of Yangtze River from School of Computer Science, Zhejiang University, is among the35 winners.

Zhou is easy to recognize. He usually stands out for his particularly big head from the crowd. Back to college time, a memoir named “My College Years” describing the legend about his “big head” and “talent” has spread on the Internet.

“His name is Kun Zhou, very young, and has strong research capacity. In accordance with Zhejiang University graduate doctoral thesis requirements, he would have got PhD the day he enrolled Master. Zhou’s head is significantly larger, with an estimated cranial capacity as at least 1.5 times as mine.” said senior Rui Lin.

“Move”

Zhou stepped into the profound field of computer graphics when he was a junior. How to make computer-generated images lifelike just as the real world? Perhaps everyone who has played computer games ever conceived it, the content of graphics research.

The 3D film “Avatar” is a typical case of graphics results, of which 60% of the images are computer-generated. However, it is a “heavy industry” which is time-consuming, energy-intensive, and far from perfect. A huge “render farm” was adopted, using 4000 Hewlett-Packard servers and 35000 CPUs to complete the film's image rendering. The total computing power of “render farm” ranks top 200 among the global supercomputers. Zhou supposed, “In use of GPU 'farm', the rendering speed will be 10 times or more, with lower energy consumption at the same time.”

Although modern GPU has complex calculations ability of rendering in HD, complex programming design and large-scale software development on the GPU is still painful for the lack of GPU programming language and debugging tools. In 2008, Zhou resigned his research director from Microsoft Research Asia. As one of the ninth Scholars of Yangtze River, he returned to his alma mater Zhejiang University and has been working for State Key Laboratory of CAD& CG since then. “University has a free research climate allowing us to do what we want.” Focusing on GPU-computing, Zhou and his team established a GPU parallel computing platform for the software development of GPU. The platform included high-level GPU programming language, compiler and debugging system, and greatly improved the efficiency of GPU software development. On the basis of prior work, they developed HD rendering software named “Render Ants”. The results were published in the top journal “ACM Transactions on Graphics”, and caused a strong reaction in both academia and industry.

This work means to move the overload of CPU to GPU, making the voluminous task quicker and easier. Insider exclaimed: Render Ants = Render Man + GPU!

“Ants” challenging “Superman”

“Render Man” was used as the image rendering software for “Avatar”. It is based on CPU, designed by the famous American animation studio Pixar and widely used in industry. Like the famous Spider-Man and Iron Man, Render Man has the symbol of American heroism with a “Man” suffix. When naming their new software, Zhou changed “Man” to “Ants”: “GPU contains hundreds of cores. Each core’s computing power is limited, but the capacity of parallel operation is powerful. The speed of Render Ants is 10 times faster than Render Man achieving the same rendering effect in the test.”

The demonstration and application of Render Ants was conducted in a national animation company. The manager had a background of administration, but was also keen on new technology. He was congenial with Zhou. Initially, Zhou just helped the company to complete some individual improvements in technology. As the “Ants” software became more mature, the company began to use it as the 3D animation rendering system in 2011.

“From modeling to rendering, the current high-level graphics processing software is almost monopolized by foreign companies. Researching on graphics for many years, we finally have our own software. We are convinced to break the monopoly. “Zhou was gratified at this.

One of the Thirty-five

“Technology Review” of MIT commented Zhou's work like this, “Thanks to Kun Zhou, computer games will become more realistic and animated movies will reach cinemas faster.” Every year” Global Outstanding Young Innovator” award, conducted by MIT, selects the scientific elites who most probably change the future, from more than 300 nominees under the age of 35 working in energy, medicine, computing, communications, nanotechnology and other fields.

Zhou said his original research intention was not to develop software, but being engaged in basic research and publishing high-level papers. He has published more than 30 papers in the top journal “ACM Transactions on Graphics”. In 2009, he was also invited to serve as the first Chinese Editor of the journal for” ACM Transactions on Graphics”. However, no matter how excellent the paper is, it would only be appreciated by academic peers. If we want more people to benefit from technology, we should use it to change our life and turn the basic research results into application. With this faith, Zhou and his team began to combine the basic research and the applied research.

Zhou was awarded” Outstanding Young Innovator of the world” for the good application prospects of “Render Ants”. Among the winners over years, we see Face book’s founder, Google’s founder, and other influential name. They are the best annotates of “Outstanding Young Innovator of the world”.

Zhou thought Render Ants had many opportunities in face of the booming domestic animation industry. “America's animation industry is very mature, and companies have strong research capability. University researchers only need to work out a single algorithm, and then the companies can integrate these algorithms into their products. In contrast, the animation research capability of China is weak, and a single algorithm can’t be applied by company directly. Companies need a complete solution. For university researchers, in order to make the new technology into application, we have to develop platforms and systems and provide complete solutions. Of course, in turn, it is helpful for the research work. We know what the industry needs and what’s the most important, so a better job can be done.”

International competition

“With the success of rendering, the application of GPU parallel computing platform will thrive in the physical simulation, massive data analysis, biological information processing and other compute-intensive areas.” Today, Zhou has a team with 20 researchers, and their research runs through three levels of GPU parallel computing concept, namely, platform, interface, and application. “Basing on our GPU parallel computing platform, the interfaces are open for other researchers to do GPU applications, and we conduct some key GPU applications ourselves.” Thanks to the idea and the mode, Zhou’s team has a unique advantage in the field of GPU parallel computing research, and has kept international advanced.

Students in laboratory all know legend of this young doctorial tutor. In 2005, Zhou created a record of the top international computer graphics conference ACM SIGGRAPH, publishing three papers as the first author. He kept awake for 72 hours before submission. Time was cut into three so that he could switch in the three different research projects. “You must believe you can do as good as the researchers in MIT or Stanford, and you must think so. Though we are separated by vast oceans, the opponent is out there, working hard. In order to participate in international competition, young people should have self-confidence and a sense of urgency, and keep fighting. "