AI Media Production Lab
In an IBC Accelerators first, the 2024 Programme will feature three very special challenges under an umbrella theme of ‘AI Media Production Lab’, featuring an all-star cast of Champions and Participants set out to explore a series of specific AI concepts in various production scenarios within this exciting, fast-moving and disruptive area of media and technology.
This will encompass researching and developing learnings around various AI tools and techniques to improve creativity and efficiency in production workflows, stamping out bias and AI hallucinations to embrace audience sentiments in content development, and exploring predictive video workflows in sports production,
Supported by Champions: RAI, EBU, Zwart, BNNVARA, YLE, Evangelische Omroep (EO), Channel 4, Paramount Global, Verizon, World Freestyle Football Association, ITV, Vodafone, Accedo
Supported by Participants: Pluxbox, Plan IX Labs, Respeecher, Xansr Software Private Limited, RKG Group, Somersault, Magnifi
The three Case Studies within IBC’s AI MEDIA PRODUCTION LAB include:
Generative AI in Action
The Challenge:
Generative AI will revolutionise broadcasters' workflows by automating content creation, enhancing real-time customisation and streamlining production processes, thereby driving efficiency and innovation in the dynamic media broadcasting industry. This challenge is about using Generative AI for content production and explore the integration and impact of Generative Artificial Intelligence (Generative AI) in a production centre environment.
As the M&E industries strive for increased efficiency, automation, and innovation, Generative AI emerges as a powerful tool with the potential to transform traditional production workflows, so the project will also aim to deliver an in-depth analysis of existing tools on Generative AI and its best applications in the creation of scene, script and virtual character.
In addition, the R&D will explore the qualitative aspects of the human-machine collaboration within the production centre, and investigate the interaction dynamics, employee skill adaptation, and the overall organizational culture in the context of introducing Generative AI into the production workflow. The POC results will aim to provide a synthesis of findings, offering insights into the transformative potential of Generative AI in a production centre setting.
The POC Objectives:
-
Create and demonstrate unique workflows utilising a variation of AI tools, that will represent pre-production, production and post-production
-
Use these POC workflows to create a fully animated pilot, inspired by the characters and storyline of ‘Around the World in 80 Days’
-
Document best practice learnings with these various AI tools, especially towards 3D models, scriptwriting, speech and world building
Watch the Generative AI in Action Challenge Pitch:
AI Audience Validation Assistant (AAVA)
The POC Objectives:
-
Create a prototype persona that will promote accurate representation and diversity in media content production
-
Create metrics to show loyalty of media consumers
-
Show the urgency of changing dynamics in the landscape and media consumption
-
Proposed POC solutions to explore interactive feedback loop with target audience during content creation
Watch the AI Audience Validation Assistant (AAVA) Challenge Pitch:
The Challenge:
This unique challenge has a core focus aimed at the entire industry and audience perceptions, utilising AI as a tool to combat bias detection, AI hallucinations and promote inclusion and diversity for audiences and focus groups at early content production and development stages.
Understanding and engaging the target audience is paramount for media organizations, so they can produce content that resonates with all parts of that audience. Traditionally, success was only visible post publication or through costly and time-consuming methods like focus groups. In an age where data is key, this project leads in innovation by combining extensive audience data with advanced AI to blend and transform content creation, involving the audience in the development process.
The project will explore solutions to turn passive viewers into active contributors through AI-driven personas that reflect society's complexity. This unique approach accelerates content development, infusing it with genuine societal insights. Essentially, the POC results will showcase a vibrant, interactive platform where the public not only consumes media but also plays a role in its creation, heralding a new era of public-engaged media.
Changing the Game: Predictive Generative AI
The Challenge:
Proposed by Champion Verizon, this case study will look at AI as a production tool for Live Sports and Live Events to assist directors and production teams to potentially predict next frames and next moves for swifter edits, content detection, and better workflows. Leveraging LLMs, tailored datasets and computer vision housed in a Private 5G network, this Challenge aims to create the ability to generate predictive video and audio that is authentic and as close to what actually happens during a live-action event such as sports or live broadcasts. The project will aim to evaluate the efficacy of which datasets are needed to accurately predict live video outcomes and how far into the future the predictive video and audio can remain authentic to the actual video and audio, to help enable real-time enhanced experiences, such as personalized commentators and localization.
This Challenge POC results will also be mix of new exploration in LLM and Modelling, Computer Vision and Generative Video (ie. Stability AI) to help the M&E industry envision new workflows to elevate live-to-air storytelling and broadcasting, and truly enhance the fan’s experience.
The POC Objectives:
-
Utilise bespoke content from one or more professional sporting competitions, and create an authentic AI conversation via end use data points, to help determine preferences for a personalised experience
-
Showcase Authenticity Calibration via a short demo clip with user preferences
-
Personalise a highlight reel of game/sport which incorporates the user's preferences to include multiple personalized elements, for example:
- Commentator language/tone/style preference and interactive/reactive
- Type of video content shown
- Overlays of stats of the team or specific players
- Subtitles/captions
Watch the Changing the Game: Predictive Generative AI Challenge Pitch:
)
){kind=logo}
){kind=logo}
){kind=logo}
)
){kind=logo}
){kind=logo}
)
