AI is taking centre stage in today’s film making

When watching a film, you may be the sort of person who immerses themselves in the story and special effects with a view to a couple of hours of escapism. Or, perhaps like me, you are the type who wants to work out what is real and what is computer generated imagery (CGI) and how realistic it really is. Either way, filmmakers continue to push the boundaries to improve the quality and variety of the special effects they deliver with the purpose of enhancing the audience experience and keep us coming back to the box-office. Films are now leaving the studio and the location shoot and moving in a steady stream towards the data center. The latest wave of technology seeing adoption includes areas such as Machine Learning and Deep Learning, which are all subcategories of artificial intelligence (AI).

Machine Learning has already made good progress in the world of film...and it isn’t all about recasting Nicolas Cage in every role (though there is an App for that)! Streaming services already use Machine Learning to recommend new content based on your preferences, going beyond the capabilities of traditional recommendation engines. Behind the scenes, algorithms are also used to support many of the processes involved in pre and post production. For example, the use of cognitive data science has been widespread in the process of pre-screening potential scripts to accelerate the process of finding good stories suitable to go into production - one great example being the service offered by ScriptBook. Video content reviewing techniques are also used on completed sequences to determine which clips might be most suited for use in the film trailer with the aim of maximising audience figures.

AI has of course also been used to enhance the quality of digitally animated characters such as the swarms of Zombies in World War Z. It is also beginning to tackle many more routine tasks like smoothing out special effects; as well as a few more complex ones such as adding realism to a CGI character by enhancing bodily movement. At the very high end, this may also include difficult and subtle tasks such as improving facial expressions - check out Serbian company 3Lateral who deliver amazing work on facial morphing.

The benefits of AI in these endeavours is that they free up the time of the artist to concentrate on the more creative aspects of their work. Data Scientists are tasked with automating the repetitive activities, freeing the artist to unleash their muse and with the time saved produce better films.

Still, AI does have its limitations for the sector, though over time many of these may be addressed by a newer form of AI called Deep Learning that tries to approximate processes in the human brain. So while Machine Learning can be used to help support the creation of engaging characters and worlds, overall AI hasn’t yet progressed to the point where scriptwriters or actors can be replaced, and some will argue that is unlikely to ever happen given the advantages a human brain has over an artificial one.

AI is not yet ready to create engaging and exciting stories that will achieve emotional engagement from an audience and similarly, it lacks the subtleties and understanding of the human condition that is the key to many great acting performances, plots and scripts. While we may see increasing use of CGI and much more photo-realistic images, for the time being, these CGI characters will still be driven from behind the scenes by talented writers and actors playing an alter-ego.
Underpinning these current and emerging capabilities are ever more powerful technology, both hardware, and software, the latter of which will be further enhanced over time by Deep Learning and Neural Networks trained with copious quantities of data.

As a provider for industrial scale Deep Learning and Machine Learning platforms and environments, Verne Global is ideally positioned to support film and media organisations leveraging the latest in AI technology. Not only is Verne Global’s infrastructure uniquely optimised for the most advanced forms of Deep Learning, but we also support the high-performance graphics processing (GPUs) as well as more traditional processors (CPU) infrastructure needed for the most demanding rendering requirements. Our specialist platform - hpcDIRECT also includes the ability to pre-install and optimise the latest rendering applications such as MaxonCinema4D, Autodesk Maya, Autodesk 3ds Max with associated plug-in’s. To deliver the specialist application services, Verne Global is delighted to be working closely with its partner RenderNation, who has over 15 years of experience in advanced rendering services to the film, television, game and architectural sector.

All of this capability is underpinned by Verne Global’s Icelandic data center, which offers low-cost green energy from geothermal and hydro-electric sources, along with free air cooling (vital for keeping all of this power-hungry IT equipment cool). So not only is Iceland the ideal location for filming with its stunning natural views such as waterfalls, glaciers, moonlike volcanic landscape and 21 hours of daylight during the summer, but it is also the ideal location for your compute-intensive post-production services, especially those levering AI. This is why Verne Global’s services have been used in films such as Contraband, 2 Guns, and Everest to name but a few. The Iceland Film Commission is also very welcoming to filmmakers and even offers a 25% rebate on production costs incurred in the country.

If you’re working within the graphics, rendering, or pre-and-post production industries and looking for a highly scalable, low-cost solution to use for your latest compute intensive film project, let us know. Verne Global, together with the specialist support from RenderNation are here to help.

Reproduced from my original blog at: https://verneglobal.com/blog/ai-is-taking-centre-stage-in-todays-film-making

Vasilis Kapsalis is the Director of Machine Learning and HPC at Verne Global