中文


Computational Aesthetics


Dr Kang Zhang is a Professor of Computational Media and Arts at Hong Kong University of Science and Technology (Guangzhou), Professor Emeritus in Computer Science at the University of Texas at Dallas, and Guest Professor of China Acadamy of Arts. He and his team invented and developed a new generative art software system, known as GALA (Generative Art for Limitless Applications), capable of automatically generating abstract images of Bauhaus and well-known artists styles, including Kandinsky, Klee, Miro, Mondrian, Picasso, and Pollock. The techniques and algorithms used in the system have been published in SIGGRAPH Asia, Leonardo (MIT Press), The Visual Computer (Springer), IEEE Transactions on Visualization, and Computer Graphics, and other international conferences.

In collaboration with Professors Kaiping Peng and X.J. Sam Zheng of Tsinghua University Department of Psychology, Professor Zhang has developed PsychArt, a personalized image generation Web system based on the Big Five Personality Traits. PsychArt provides a user-friendly interface that allows the user to select the scales of ten self-descriptive items and automatically generate a Kandinsky styled image representing the users personal profile. It was first experimented at Alibaba's Hupan University in China with great success.

The abstract images and design generated by GALA have the following advantages, that maximize the flexibility of personalization and customization:

  • No copyright issues
  • Millions of unique images could be generated in seconds or minutes
  • All parameters (colors, sizes, composition, etc.) are adjustable and randomizable
  • Different artistic styles could be combined and integrated
  • Any additional features (deco, logo, text label, signature, etc.) could be inserted anywhere as desired
  • Using the Kandinsky styled visual elements, Professor Zhang has also developed a parameterized generator for Kandinsky Fonts, including 26 capital case and 26 lower case letters.

    Example products using the generated images include silk scarves, ties, mugs, and mousepads, as shown below. One application of the generative techniques is to generate abstract images of desired styles for hotel and restaurant decoration and interior design. A typical design is to assign a single style of distinct paintings to each functional area or floor, including corridors and rooms (as modeled in the following figures). The cost of generating and producing these paintings is minimal comparing with manual processing or painting.

    Click HERE for art / design related publications, HERE for computer science publications.

    In his spare time, Kang likes to swim, play badminton, travel (see his travel page), and play go, chess, and bridge. Oil painting (won 3rd Prize in Shanghai in 1984. His recent works), paper art, logo design, porcelain painting, and Chinese calligraphy (won 1st Prize in Sichuan Province in 1981. The style Kang learnt is Liu Gongquan) (his own works in Regular Script, Clerical Script1, and Clerical Script2), and photography are his real hobbies.


    (Co-)authored or (co-)edited books


    A few example products made of GALA-generated abstract images:

    A university classroom decorated with Pollock styled image (real):

    Hotel interior design (abstract art images pasted):

    Selected screenshots of "Mind Painting" (first used at Alibaba Hupan University in September 2015):