Pre-Visualization: From Picture Continuity to High Quality Movie
Pre-Visualization: From Picture Continuity to High Quality Movie
  • Reporter Lee Ji-a
  • 승인 2015.03.18 09:41
  • 댓글 0
이 기사를 공유합니다

In “X-Men 2” characters can jump high and have supernatural power. Acts are too different than what people can do in life and the director have to depict more details in actors. So the director produced continuity in video. It takes seven months to make a forty-five-minute video and a scene of just a few seconds costs 1,000 - 2,000 KRW. This becomes extremely expensive. To solve the problem, Professor Jehee Lee’s (SNU) team developed an algorithm that makes pre-visualization easy. The Postech Times interviewed Jungdam Won (SNU), who has researched this program for one year under Prof. Lee. (The interview was made by phone and the order of some content has been rearranged.)

Q. How was this pre-visualization technology developed?
To produce a movie or performance it is essential to discuss how and where actors move in each scene. Seeing a direct simulation is helpful and there are some ways of pre-visualization. Simple handmade images are common and in case of big company they use simple key frame animation technology. But depictions are abstract so depicting interactions of many characters requires pre visualization artists, which increases the cost and production time.

Q. How does this technology work?
After inserting characters’ names in boxes of the program and lining up verb boxes with suitable characters, the program automatically generates video. For example, there are “Tom”, “John”, and “David” boxes. The boxes “Tom” and “John” line up with “Bump” and “Tom” box lines up with “Point” and “Kick”. The “David” box then lines up with “Join” and “Fight”. After that, the program creates a video in which Tom and John bump into each other, Tom points at John and kicks him, and David joins and fights. Besides the video, the program also shows other candidates. There are many ways to fight. For example, characters can resort to force or kick, and the scene becomes different depending on where one character hits another and the variation of movement. So a word has various scenes over two. Nearly countless sequences of video can be produced. When people search for specific information in Google, Google shows the information order, the most relevant items are displayed at the top. Google uses a ranking algorithm, and the research team expanded it to animation. It recommends some representative scenes and the director can choose the most suitable scene. Won said “Producing many scenes is very complicated. But our algorithm determines the probability of movement will be satisfied because every movement from characters lines up accurately temporally and spatially. Also, the director can decide strength of hitting or kicking by regulating kinetic energy. When the script or director’s preference is changed, the program quickly provides other candidates and it improves the efficiency. If directors want additional actions, they can add specific actions to the database. Actors attach sensors to themselves and act based on script. Then markers pick up the actions of the new movement, which is later included in a database. The more information added to the database, the better this program will become.

Q. What is the effect?
As mentioned above, the technology can save a lot of time and money. And directors can see a simulation directly and make better choices about details more quickly. It will improve the quality of movies. Won said, “We introduced the automated program instead of hand made and it accelerates communication between directors and actors. Finally it improves efficiency of producing performance and decreases cost.” For these reasons people expect this to help the movie industry develop.

Q. What more is needed and in the Future?
Normally it is rare for a fight to occur strictly between people, without the use of or interaction with objects. People throw and brandish something, or people lean against a table or wall. The technology does not include objects. We still need technology to project the interaction between people and objects. Won said, “At this point, do not plan to research this area. It will be very complicated and difficult. We expect other research teams will successfully develop this technology.”
Even though the technology is not perfect, it can be a chance to produce video on an unprecedentedly large scale. In the future, technology will create videos directly from a script and one or two directors will be able to make movies with the same quality as now in a week. Finally, this type of technology will lead to a new era of personal movie production.