T+T Icons In Science & Technology Volume 3
Anyone who has not heard of Anil Kokaram will discover, within seconds of searching for him online, one striking fact: he is an Academy Award winner. Clearly it is hard to tell his story without rushing to mention this. After all, the Hollywood mystique is compelling. But there is an added charm in Kokaram’s case because most people do not associate scientists with those red carpets.
Yet as far back as 1931, the Academy of Motion Picture Arts and Sciences has recognised many scientists and engineers for their contribution to advancements and milestones in the development of technology for the motion picture industry. In 2007, Kokaram became the first and, to date, only Trinidad and Tobago national to win an Academy Award. Currently Research Manager at YouTube/Google’s Video Processing Group, Kokaram is an internationally recognised master in the field of digital signal processing, and particularly motion estimation. He and his colleagues, Dr. Bill Collis, Simon Robinson and Ben Kent at The Foundry, one of the world’s top software development companies for post-production visual effects, were awarded a Scientific and Engineering Award for their breakthrough work on motion-based special effects for the film industry. The FURNACE program they developed was a truly revolutionary compositing and editing software suite for video processing in post-production. It allowed the user to perform automatically tasks which were not possible previously even with manual editing. By harnessing Kokaram’s ideas about making ‘pictures between pictures’ i.e. creating new frames, users can add, remove and manipulate characters and objects in scenes, remove ‘dirt’ and even restore old footage.
The effects in The Matrix for example, are not possible without motion estimation. Actors are surrounded by a ring of cameras which record at the same time, so that at any instant a picture is taken from many different points of view. It is not possible to place the cameras together close enough to make the “Matrix effect” look smooth. The smooth effect is creates by synthesising new pictures as if they came from cameras that were virtually “in between” the ones actually there. To build these new frames requires estimating the motion between existing frames. Kokaram developed algorithms to solve problems like these and the Foundry team was able to build those ideas into FURNACE.
At the time of the award, FURNACE had been used in a host of blockbuster films including The Matrix, Casino Royale, X-Men 3: The Last Stand, The Da Vinci Code, Charlie and the Chocolate Factory, Batman Begins, King Kong, The Lord of the Rings and many others.
Behind the glamour, the Academy’s science and engineering nominees are subjected to rigorous scrunity during the assessment by two panels of judges made up of academics and representatives from the film industry. “Almost like a police investigation,” Kokaram jokes. But while he says he was surprised to have won, others who know him well may not be. Kokaram has always been an outstanding achiever, whose life has been studded with scholarships and prizes.
He was born in Sangre Grande, Trinidad in 1967 to Richard Kokaram, a former principal of Hillview College, and his wife Lynette, also an educator and a past principal of Tacarigua Presbyterian School. The family relocated to Curepe, where Anil and his three siblings grew up. He has particularly fond memories of their many family vacations in his father’s hometown of Fyzabad.
He was intrigued by science from an early age and, with his boundless natural curiosity, found many avenues to explore and stoke this interest, which was greatly encouraged by his parents. He became a big fan of the documentaries of astrophysicist, Carl Sagan, and naturalist, Sir David Attenborough, and was struck by the realisation that it was possible to make a career – and on TV! – out of research. He was an avid reader of science fiction and also immersed himself in his home chemistry set.
His passion for experimenting grew as he left his primary school, Curepe Presbyterian, for Hillview College, where he made new friends also interested in science. At 13, with one of these friends, he took up the hobby of building model rockets and aircraft – a surprisingly hazardous undertaking that involved jumping fences and dodging stray dogs to retrieve his models wherever they had crash landed! His desire to perfectly predict and control the flight and fall of his creations, by measuring wind speed and setting trajectories, enabled him to see more concretely that science was not only an interesting pastime, but a practical field with real-life applications, and that “physics and mathematics could solve any kind of problems – even though in practice it never worked with my rockets.”
Already enamoured with research, Kokaram performed well in science at school. Mathematics on the other hand required extra effort. He says, “For some reason I was pretty bad at math until about half way through secondary school. I even failed at one point. Around age 13, my dad had a ‘discussion’ with me about this and magically my math grades changed! I still consider myself not great at it, but I have ended up in a pretty math-heavy discipline. I think many people are turned off math because they just see numbers and abstract ‘stuff’ with no real purpose. But I was lucky in that I was able to see what it was good for and I still motivate myself to study it because I know it helps me make cool cinema effects.” He certainly did get over the hurdles as, in 1985, he won an open national scholarship, coming fourth in the science group with his A level results in mathematics, physics and chemistry, and also staying on to teach mathematics at Hillview for a year.
He then took up a Tate and Lyle scholarship to pursue electrical and information sciences at Cambridge University. His was a balanced academic life with time allotted for hard work, cricket and socialising with the many people he met through the sport. In his final year, he collaborated with a friend, Graham White, studying zoology at The University of the West Indies (UWI) to build a recognition system for frogs in Trinidad using digital recordings of their croaks, applying concepts normally used for human speech recognition.The idea was toenable scientists to automatically inventory large areas by just recording sounds in a field. This project ignited his interest in digital signal processing (DSP), a discipline that lies somewhere at the intersection of electronics, computer science and statistics. Kokaram says,“You could go directly from the mathematical analysis of a problem to a working system that interacts with the real world, all without having to actually build electronic circuits and, in fact, just by writing the ideas into some software.”
DSP originally arose from the study of improving reliability of digital communications. The foundations of the work were established during thesecond world war when Norbert Weiner, the celebrated mathematician and philosopher, and ‘father’ of cybernetics, helped to design circuits that maderadar echoes “cleaner”,and therefore made it easier to spot enemy aircraft.As Kokaram explains it, “DSP has evolved to become the study of extracting useful information from measurements. For instance, when you watch a video over the internet, a computer program is measuring the bandwidth available to you, and when that drops, it alerts the video server to switch to a different video file which will play at that lower bandwidth. The estimation of what bandwidth is available is done by applying DSP methods.”
In 1989, Kokaram graduated with his combined Bachelor of Arts/Master of Arts, Electrical and Information Engineering Sciences Tripos (with honours), winning two Churchill College Cambridge Prizes for Academic Merit along the way, and spent the next four years pursuing his doctorate in signal processing. He wrote his thesis on applying those techniques in movie restoration and correcting video defects, notably filling in holes in pictures, and though he didn’t know it at the time, that research would lay the foundation for his groundbreaking achievements later on with respect to motion estimation.
After graduation, he stayed on at Cambridge as a Research Associate and Fellow at the Signal Processing Group of Cambridge University Engineering Department. During the mid-90s, he became the first researcher to develop techniques for film restoration that successfully used a statistical construct called Markov Random Fields (a fancy name for describing the idea that small patches of pixels tend to have the same colour).This is now a common tool in post-production houses worldwide. In 1996, he published Motion Picture Restoration – Digital Algorithms for Artefact Suppression in Degraded Motion Picture Film and Video, a book on movie restoration that remains a key reference among professionals in the field to this day.
In 1998, Kokaram left Cambridge and made three decisions that would change the course of his career – he accepted the position of Lecturer at the Department of Electrical and Electronic Engineering in Trinity College Dublin; founded the College’s Signal Processing Media Applications Group (Sigmedia); and began building links between his newly founded group and The University of the West Indies (UWI), St. Augustine, sponsoring several Caribbean students to study at Trinity. In the years that followed, Sigmedia would host an array of projects, applying signal processing to digital cinema, multimedia information retrieval, video over wireless, surgery, underwater video enhancement and atomic scale microscopy. Kokaram has raised over two million euros in competitive funding for research and, since 2000, has led that team through more than five large EU projects.
During the post-production of The Matrix (which was released in 1999), someone Kokaram had met at Cambridge in 1995 was working on the movie and he approached Kokaram for DSP assistance with a problem they were having with one of the effects. It was then that Kokaram developed the new techniques for motion estimation that became an industry-changer. His successes won him further work on the Matrix sequels, and also led him to establish his own start-up motion picture restoration and video processing company in 2004 – Green Parrot Pictures.
Between 1999 and 2007, Kokaram consulted widely in the area of video processing, working with Daimier Benz and Cambridge Consultants as well as The Foundry, with whom he developed a range of new tools. This work ultimately culminated in the development of the FURNACE program with Collis, Robinson and Kent. In 2011, the Internet media giant, Google, bought Green Parrot Pictures and contracted Kokaram and his team to help the company improve the quality of the vast video archives – 72 hours of video uploaded per minute, as of 2013! – stored on their YouTube servers. His team is responsible for helping YouTube achieve a balance between compressing the video down to manageable bandwidths, and still keeping it looking great. They also work with other teams in Google. In April 2013, YouTube launched the SloMo effect in their editor which was created by Kokaram’s team.
Prof. Kokaram has published over 100 papers in signal processing. Between 2002 and 2006, he served on the editorial board of the IEEE Transactions on Image Processing, the most important journal in that field. He also continues to supervise PhD students at Trinity while maintaining links to UWI, St Augustine, where he has helped to create a new course in video processing. He has worked with NIHERST on science popularisation and takes every opportunity to speak about his field, running workshops at conferences in the UK, USA, Trinidad and Tobago and his home base of Ireland.
Along with the industrial and commercial applications of Sigmedia’s research, Kokaram also has a personal interest in applying his work to a lifelong passion – cricket, viz. creating effective automated highlights packages to fill the large demand produced by the growing number of cricket matches being televised globally.
Despite his work load, Kokaram still manages to find the time to “read a lot, watch too many old movies, play cricket and talk to people in other disciplines so I can find new problems to work on.”
He advises young people that a career in science “helps you to learn how to learn… you can always find something interesting to do. All industries are in need of engineers and scientists to design new products or analyse strange patterns in data. Students should foster a love for reading, studying and socialisation – the first two will help them appreciate the myriad new and emerging applications in science and engineering, and the last will help them discover opportunities to utilise their unique skills. Meeting people from different disciplines exposed me to other problems and sometimes I was able to help. And in the course of doing that, you invent something new.”