Darren Aronofsky & Eliza McNitt on AI, Filmmaking, and Storytelling
GoogleJune 20, 202540 min26,629 views
33 connectionsΒ·40 entities in this videoβThe Making of ANCESTRA
- π¬ The film ANCESTRA was created through a partnership between Darren Aronofsky's Primordial Soup and Google DeepMind, marking a new era of AI in filmmaking.
- π₯ The production involved a large team of 200 people, including a dedicated "AI unit" comprising writers, animators, art directors, technologists, and prompt engineers.
- π‘ Eliza McNitt emphasized the importance of human performance at the core of the story, driving the narrative about a mother and child.
Integrating Live Action and Generative AI
- π₯ Decisions on blending live-action and generated footage were critical, with live actors centering the human element and generated imagery exploring abstract concepts like the cosmos and the human soul.
- π¬ The AI unit, composed of filmmakers and artists, leveraged their skills to adapt nascent generative technology to director Eliza McNitt's vision.
- πΌοΈ Prompting for generative video involves condensing complex visual decisions into text, requiring creative problem-solving, such as using specific microscope types to achieve unique imagery.
AI's Impact on Cinema
- π Darren Aronofsky believes AI will free filmmakers' imaginations to explore more diverse and unexpected creative avenues.
- π‘ The discovery process within large AI models is exciting, with potential for storytellers to guide these tools for narrative purposes.
- π οΈ Generative AI offers new ways to create visuals, akin to Aronofsky's past use of practical photography and chemical reactions, but now through machine prompting and manipulation.
Generative AI and Human Performance
- πΆ The film features a generated baby, created from Eliza McNitt's childhood photos, highlighting AI's ability to synthesize realistic human likenesses.
- π Traditional VFX challenges, like depicting a newborn realistically, are being addressed with AI, potentially making difficult shots easier while introducing new unpredictability.
- β‘ The speed of AI iteration allows filmmakers to experiment rapidly, as seen with generating baby visuals on the second day of production, informing subsequent on-set strategies.
Future of Filmmaking with AI
- π AI tools can accelerate research and pre-visualization, enabling directors to solidify their vision and communicate it more effectively before production.
- π€ The future of filmmaking may involve a collaboration with machines, creating new workflows and job roles, but the core of storytelling remains paramount.
- π Filmmakers are encouraged to be inventors, pushing boundaries with AI, as it's likely to transform the industry in surprising ways, emphasizing that AI won't replace storytellers but change how stories are told.
Creative Control and Collaboration
- π§ The creation of a complex oner shot involved modeling the human body in VR, then using AI to generate consistent eight-second segments that were seamlessly stitched together using traditional VFX techniques.
- π The film's soundscape is deeply personal, incorporating extracted vocalizations of Eliza McNitt as a baby and recorded heartbeats of her and her mother, blending the cosmic with the intimate.
- π¨ The integration of AI tools allows for greater creative control and efficiency, enabling filmmakers to focus more on performance and artistic vision, even while embracing new technologies.
Knowledge graph40 entities Β· 33 connections
How they connect
An interactive map of every person, idea, and reference from this conversation. Hover to trace connections, click to explore.
Hover Β· drag to explore
40 entities
Chapters15 moments
Key Moments
Transcript147 segments
Full Transcript
Topics15 themes
Whatβs Discussed
Generative AIFilmmakingArtificial IntelligenceANCESTRAGoogle DeepMindDarren AronofskyEliza McNittPrompt EngineeringVisual EffectsStorytellingAI UnitPrimordial SoupVeoVirtual RealityCompositing
Smart Objects40 Β· 33 links
PeopleΒ· 12
CompaniesΒ· 3
MediasΒ· 8
ConceptsΒ· 11
ProductsΒ· 3
EventΒ· 1
LocationsΒ· 2