How a Hyderabad-based start-up is using data science to predict box office results and help filmmakers make informed decisions
Filmmaking is often a leap of faith, where no one is sure of box office results. Experience makes people wiser, but there’s also the tendency of staying in a ‘safe zone’, going by past market trends of what worked and what didn’t. Only a few risk-takers steer away from the norm. However, given their understanding of new age audiences and their ever-changing preferences, they may not really be off the mark. But, can this whole business that surrounds the art form be less of a gamble? It’s possible, with some help from big data analytics and Artificial Intelligence (AI).
Internationally, the entertainment business gave big data analytics a serious thought after Netflix leveraged it to launch its House of Cards series, predicting that it would be a hit with the viewers. Consider a similar scenario in Telugu cinema. Woodcutter Film Technologies, a firm that works from Ramanaidu Studios, Hyderabad, has been using machine learning to build a reliable tool to help filmmakers at different stages of making, promoting and releasing a film. In June 2018, Woodcutter was one of the seven start-ups selected for an accelerator programme mooted by Anthill Studio, which is run by Suresh Productions and Anthill Ventures. Their predictive analytics model was built in mid 2018. They applied it to Care of Kancharapalem and were 64% accurate in predicting box office numbers. In the next two months, they tweaked their method by adding more parameters and data. The accuracy level touched 97% for Taxiwaala (see infobox). Predictive analytics takes into account the genre of film, number of screens released, pre-release buzz, and the box office potential of the cast and the director.
Core team: Bethanavel Kuppusamy (founder, CEO), Devanand Guntur (data scientist) and Ashwin Venkatesan (data scientist)
Accuracy in box office prediction: Care of Kancharapalem: 64%, Savyasachi: 86%, Amar Akbar Anthony: 92%, Taxiwaala: 97%, Antariksham 92%, and Padi Padi Leche Manasu: 96%
What next? A computer vision solution to capture audience’s facial reactions during a film’s test screening to analyse their emotions for every scene. The research can help identify scenes that may require editing or reshooting.
Woodcutter’s partners and clients include AGS Entertainment (Chennai), Suresh Productions (Hyderabad), Excel Entertainment (Mumbai), A R Rahman, and Rana Daggubati.
At the helm of Woodcutter is its founder-CEO Bethanavel Kuppuswamy, an engineer from IIT Madras who took up theory of filmmaking as an ancillary course. A film-tech company wasn’t in his scheme of things. He wanted to make short films and eventually direct feature films. Today, he’s enjoying the learning curve in the number crunching that can help directors, actors and producers take data-driven decisions to try and make successful films. “Filmmaking is actually a science,” he tells us, when we talk about finding method to the madness of movie-making.
In late 2014, Bethanavel worked with Ola and got a taste of launching cab services in different cities. He equates it to a “start-up within a start-up” kind of experience. His interest in entrepreneurship was piqued. At the same time, as an aspiring filmmaker he discussed ideas with his friends and worked on prototypes — ranging from creating a platform to connect with and hire cast and crew, like a Linkedin for filmmakers, to envisioning an app that could ease production niggles by ensuring smooth communication between crew members, while also quantifying social media buzz for their project. “While exploring film financing and distribution trends for independent films, we learnt that the professional film industry needs technological intervention to improve efficiency,” says Bethanavel.
Woodcutter predicted the first day BO collections of ‘Taxiwaala’ with 97% accuracy
He discussed these ideas with a few studios in Chennai and Mumbai and found that filmmakers were eager to get a pulse of social media buzz. “Earlier we had one theatrical trailer prior to a film’s release. We then began having teasers, first look posters, title announcements, project announcements, and now fixing a date and time for an announcement has also become a big deal,” he points out. Machine learning helps to quantify all this social media engagements for filmmakers and accordingly plan promotional and release strategies.
In 2017, Bethanavel’s team did social media analytics for Fukhrey Returns, gauging what the audience liked in its teaser — down to the details of dialogues and which actor drew more attention in the multi-starrer.
Woodcutter gleans data from Twitter, Facebook, YouTube, websites, and other media platforms. Data also comes in from production houses and distributors on screen counts, success of past films, and more. All this helps build a formula for predictive analytics. “We cannot change a product, a film in this case, but we can improve the chance of having better return of investment,” Bethanavel explains.
Recently, Woodcutter began working with actors to analyse their popularity and fan following. They’re also working towards pre-production consulting where scripts will be analysed scene-by-scene using AI tools to get in-depth understanding of story elements, character trajectories and success probability. “Scriptwriting is an art that has structure and grammar. And technology can be used to help scriptwriters get these fundamentals right and also take learnings from past film scripts. From our discussions with new age filmmakers in India and production houses in Hollywood, we’re able to see this is a pain-point for the global filmmaker community and needs to be addressed,” Bethanavel emphasises.
Using machine learning to the art of writing a script may seem peculiar, but Bethanavel states that directors Woodcutter has spoken to are excited about the possibilities: “Because they adhere to thumb rules and know that the writing process is a science,” he signs off, adding that on a personal note, he hopes to realise his childhood dream of making movies a reality.
First Published On: https://www.thehindu.com/entertainment/movies/woodcutter-film-technologies-a-hyderabad-based-start-up-is-using-data-science-to-make-filmmaking-more-viable/article25931252.ece