The aim of this article is to show the integration and use of AI in filmmaking. The writer
believes without this practical applicability, AI research is irrelevant and looks at the use of AI in various production stages.
Pre-production:
The author discusses how AI can be used to collect and assess data in order to advise projects which can be greenlit for production. Using ML techniques, Netflix mastered this activity and created ‘taste communities’, clusters ’or ‘verticals grouped together on taste rather than age, race or other demographics’. This allows Netflix and other production companies to determine what people are interested in watching and expected revenue before a decision is made to produce a project or to advise the type of production to be undertaken.
Hollywood took this on with the use of AI software including Cinelytic, Vault and “Merlin Video”.
He notes that despite the ability of AI to make recommendations, the final decisions rest with humans, another clear indication that humans are still important in the production process and AI tools are tools no matter what their abilities may be.
An interesting discussion revolves around scriptwriting where AI can provide ‘hindsight’ data analysis of scripts which have worked which then provides ‘foresight’ into what needs to be done to create and leads to ‘insight’ as to what would be successful for audiences.
Interesting to note that due to scriptwriting’s formal structures such as location, genre, cast and so on it makes itself susceptible to replicate by AI.
A discussion takes place on AI scripting software called ‘scriptbook’ which seeks to collaborate or generate scripts on its own taking into consideration all the elements around traditional scriptwriting with a focus on constantly updating its capabilities to meet human standards, whilst also providing unique ways of storytelling which he feels will enhance human scriptwriting.
Additionally, it is noted that since AI uses ideas from various sources it is not inherently authentic, thus it is still the human who decides what the next steps of the project will be.
Production
AI has been used in virtual filmmaking by teaching it the film language called ‘Declarative Camera Control Language (DCCL)’. He further discusses how film production techniques are ‘replicated within virtual environments’ which has now led to the ‘Director’s Lens’ which involves pre-visualisation of shoots in collaboration with AI. There are various uses of these, for example, in shot lists, set design, camera angles and so on.
The author discusses the use of “Automated Cinematography with Unmanned Aerial Vehicle’ which use text-based input know as ‘Prose Storyboard Language (PSL)’ to film.
Thus, AI can create shots based on prompts which may influence the film produced or be used in animation as content in short films.
He suggests that films influenced by AI can be referred to as ‘creatively influenced by AI’.
Using ‘Mario’, it was proved that screenplays can be converted to short films using a high level of prompt language.
Using these AI prompts and films as precursors to advise live shoots across all areas of production.
One of the tools developed to automate Single Operator Mixing Application (SOMA) is the “Ed” AI system which allows the use of AI to automate the selection, sequencing and framing of shots for the video input which formally all used to require human input.
Post-production
As AI capabilities allow for more accessible, cheaper, quicker real time postproduction opportunities, they also enable rotoscoping which enable the generation of fine-tuned details such as dust or strands of hair.
The author discusses GAN which are now common place across scriptwriting image creation and video and the use of ‘Neural Style Transfer’ (NST) which allows for the creation of sequences in a short film. These which are now advanced techniques in film production as seen with Sora from OpenAI and Gen 3 Runway.
The article suggests that the capability and capacity to impact VFX democratises access and allows for people and companies with smaller budgets to use the software to create content.
The article’s conclusions dwells on the issues of democratisation of AI in VFX and access for smaller creatives, the ability of AI to assist with green lighting projects, and supporting AI and human collaboration. This he understands will bring disruption as the industry adapts.
Conclusion
I have just completed reviewing an article by Alex Frohlick entitled 'Artificial Intelligence and Contemporary Film Production: A Preliminary Survey'
The usefulness of this articles is its practical application of the AI tools in all aspects of the production process and how the tools impact across the entire life cycle of filmmaking.
Additionally, it concludes with an understanding of the disruption which AI will cause and yet with the optimism of the creative potential of AI-human collaboration for creative production. This disruption has been highlighted by most of the rhetoric around AI especially with job loss but a look at AI’s uses as presented in this literature indicates the development of new skills and new potential roles and jobs. It is already estimated that AI will create 700 new roles. It's in one of my other reviews--I can't remember where--check my blogs to find out)
Looking at the article written fours year ago, AI technology in filmmaking has already surpassed much of its capabilities at that time, with the capacity and promise of being able to continue to disrupt traditional methods of production with more innovative, creative opportunities. Gen 3 Runway and Sora were not available when this article was written and yet they present a clear indication of disruption to filmmaking for the long term.
The writer’s simple attribution to AI-Human collaboration including attributing ownership to AI when used provides an opportunity to discuss issues around copyright and ownership, a debate currently raging across most literature on AI in creative practice.
With the current capacity of AI tools in filmmaking-- pre-production, in production and post-production, it is imperative that academics provide these opportunities to the new generation of filmmakers so they do not leave with universities degrees which are inadequate within an AI driven industry and world.
Additionally policy makers must take the lead in creating policy around the rules governing the use of AI and must do so in collaboration with the creative industry.
Missing from the article is a look at the impact of these AI tools on people across the filmmaking life cycle. There is thus a need for further research on the impact of AI on the creative sector including audiences.
Alex Frohlick, Artificial Intelligence and Contemporary Film Production: A Preliminary Survey, July 2020, Goldsmiths, University of London
Comentarios