2025 Accelerator Project

Final Showcase Session

The Challenge:

Champions: VRT, RAI, YLE, EBU, ITV, Institute of Cultural and Creative industries (University of Kent), GLOBO, Channel 4

Participants: Pluxbox, Yaara Marchiano, Studio Deussen, Animatix, Google Cloud, AMD/HP


This project aims to develop a Generative AI Framework that enables broadcasters to efficiently create various types of media content, including scripts, highlights, short-form promos, TV ads, and marketing materials. By leveraging generative AI models, the framework will allow users to input prompts, data, and assets, generating high-quality, customized content in minutes. The goal is to streamline and enhance content production workflows, making them faster, more scalable, and more adaptive to audience needs. 

Despite the growing interest in generative AI, broadcasters face challenges in fully integrating the technology into existing production pipelines. This project will provide a clear and practical framework, outlining both the capabilities and limitations of AI-driven content creation. It will explore automation in marketing, social media, and promotional content, including dynamic ads that adapt in real time based on user interactions, making advertisements more engaging and relevant. 

Joining forces with a truly international team of broadcaster Champions in Globo from Brazil, YLE from Finland, ITV from UK, EBU and more, this initiative will further investigate AI’s role in animation production, incorporating narrative, artistic, and compositional elements to create emotionally impactful content. A key objective is to develop a 20-minute AI-generated animation, ensuring consistency, quality, and cultural relevance, particularly focusing on a strong Brazilian identity.  

Additionally, the project will analyze traditional content production workflows, assessing AI tools for smoother integration into pre-production, scripting, production, and post-production processes, ultimately helping broadcasters adopt AI in a structured and effective way. 


Kickstart Pitch:

POC Results

To achieve this, this POC AI Framework was purpose-built as a carefully customised orchestrator, and experimented with over 20 new and existing Gen AI APIs and tools, including Google, ComfyUI and more. (see POC materials below for visual architectures and ‘making of’ videos)

The team remained true to their original creative challenge theme of ‘ancient Rome’ and made an advancement from their Kickstart Day proposal to re-scope from short-form AI content made at speed (eg: ads, promos), and kept focus on their other key objectives of ‘quality over quantity’ development. This resulted in the creation of an original IP mini-series, ‘Echos of Rome’, which incorporated the AI Framework with a human-centric, compassionate narrative, characters, plot and ethical and compositional elements into an emotionally impactful photorealistic video trailer, and episodic breakdowns. 

The POC focussed on agile R&D, built with two distinct POC workflows. The first explored how instant AI creation can benefit ‘rapid prototyping’ for quick-fire storyboarding, scripting, character development, iteration and ideation. The second embraced a more thought-provoking ‘movie generator’ workflow that harnessed traditional, artistic storytelling techniques and guided the AI Framework to create an emotive, human-centric narrative. 

Early access was granted to a variety of new and yet-to-be-released Generative AI tools, such as Google’s Veo3 to test how prompts, data and original assets could result in customised, high-quality output. The team also experimented with new, cutting-edge compute power via AMD cards and HP hardware, and gained R&D learnings into how the speed of advanced NPUs/GPUs and CPUs can affect content creation, but also how a production can manage sustainable AI productions.

Another POC priority focussed on the extensive R&D and workshops on the ethics and responsibility using Generative AI for broadcasting and media production, and the impact on the future of our creative and industry skills. This in-depth white paper with descriptive visual learnings can be downloaded in the PDF below.


Key Goals & Achievements

  • Faster, Cost-Effective Production: The AI Framework dramatically reduced the time and resources required to bring original IP ideas to life, and enabled the team to stay true to human-centric storytelling to show that quality over quantity is still important. We generated over 130 minutes of video to produce 4 final minutes for the Accelerator trailer video, this gave a production ratio of around 30:1.

  • Enhanced Scalability: The AI Framework orchestrator enabled the team to produce and experiment with more content and resources to create original material, think about audience demands, and stay true to a broadcaster’s strict criteria for QC and output.

  • Greater Adaptability & Responsibility: This AI Framework allowed this team to quickly respond to market and production trends with agile content workflows, and acknowledge the important impact towards industry-recognised challenges for inclusive, non-bias, ethical and responsible development and output.

  • Human-centric approach: Because autonomous generation risks being generic or missing the mark, the team retained ultimate creative authority and placed the human creator in full control to guide the Framework requirements and content creation with specific aesthetic and narrative directions to ensure the final output met their vision.

  • Granular Creative Control: As a key feature of the Framework, the team successfully maintained control by integrating traditional, cinematic filmmaking methods (eg: camera movements, depth of field, lighting setups, focus pulls, etc) to produce precision scenes and a trailer for the POC. 

  • Asset, IP and Data Management: This POC also produced R&D for asset, API and data management needed to track the provenance, usage rights and source data for every generated scene. With a structured system, studios and broadcasters can control the sheer volume of assets, to make collaboration, revision, and compliance possible.


POC Results Showed at IBC2025

Live demos were showcased at the IBC Accelerator Zone, and a full-fledged presentation of the POC results, creative process and technical workflows were presented on stage during their Final Showcase Session.


What’s Next?

Several Champions and Participants will continue this R&D for AI Framework tools and creative production in various European consortiums, and via new commercial partnerships formed out of this 2025 Challenge.

Champions:

Participants: