Crafting With Code: How AI Storyboarding Research Shapes the Future of Creative Ideation
As part of the AI in Creative Industries team, Breda University of Applied Sciences has been exploring one of the most persistent bottlenecks in visual storytelling: the early ideation and concept development stage. Their research, which has been explored by Suryashree Aniyan in this Knowledge Gem, focuses on a key question facing creative teams today.
How can small teams, students and early-stage creators visualise ideas quickly and consistently without slowing down the creative process?
Their work on an AI-assisted storyboarding prototype offers a compelling answer. Rather than replacing artists, the tool accelerates the exploration period, helping creators move from ideas to visual possibilities in minutes instead of days.
This aligns directly with AICI’s mission: supporting ethical, human-led innovation across Europe’s Creative Digital Industry.
Why Storyboarding Needed Rethinking
Storyboards rarely receive the spotlight, yet they are the backbone of film, animation and game production. In the earliest stages, teams often struggle with three barriers.
• Not everyone can draw or visualise ideas clearly
• Iterating many variations is time consuming
• Early misalignment can slow the whole production pipeline
Breda’s research addresses these challenges by designing a tool that supports creators during the brainstorming phase. Not by making finished artwork, but by providing fast, consistent, adjustable visual references that teams can refine with their own artistic judgement.
This approach reflects a broader shift across the creative industries: AI as an enabler, not a replacement.
A Technical Approach Rooted in Creative Needs
Instead of building a single feature, Breda explored an entire toolchain of AI processes that together mimic early storyboard workflows.
1. Generating Locations
Using the Flux model, the team chose a simple line-art style that resembles traditional storyboards. This ensured that generated scenes remained practical, not overly detailed or hyper-realistic.
2. Building Consistent Characters
Consistency is one of the hardest challenges in AI image generation. Breda discovered that maintaining a fixed seed, combined with sketch-style prompts, allowed for reliable character variations in different poses.
3. Intelligent Background Removal
A bright red background was used during character generation so that models could separate the figure cleanly. This simple insight dramatically improved post-processing.
4. Character Placement Through Object Detection
Microsoft’s Kosmos 2 model was used to detect objects in a scene and place characters relative to them. This made the placement semi-automatic and more natural.
5. Understanding Prompts With Prepositions
The team trained a model to understand prepositions like next to, under and behind. This allowed the system to position characters based on natural language input, bringing user intention into the workflow.
6. Camera Shots and Outpainting
Flux Fill enabled wide shots, extra wides and Dutch tilts using outpainting techniques. Complex 3D camera moves remain future possibilities, highlighting the project’s forward-looking direction.
These building blocks reflect a careful balance between technological capability and artistic purpose.
What This Means for the Creative Industries
Breda’s work highlights several broader insights relevant to AICI’s wider research.
AI can remove friction, not creativity
The tool speeds up the messy beginning phase where ideas are raw, fluid and experimental. This frees artists to focus on narrative, emotion and style.
Skill gaps can be bridged ethically
When teams lack drawing expertise, AI can help visualise ideas without outsourcing or bottlenecking progress.
Modular AI workflows are becoming the new normal
This project uses several models and processes working together. This mirrors the direction of the entire creative sector where AI, procedural tools and real-time engines blend into hybrid pipelines.
Education and training must evolve to match these toolchains
Students need to understand not only how to prompt, but how to orchestrate multiple AI processes, assess quality, and apply creative judgement.
Open experimentation leads to transferable insights
The challenges faced and solutions created provide a roadmap for other universities, studios and toolmakers exploring AI-supported workflows.
Limitations Show Where Future Innovation Lies
Breda’s team was transparent about constraints.
Certain camera angles require 3D models.
Some poses remain difficult to generate consistently.
Superposing characters inside complex environments has limits.
These limitations do not diminish the outcome. Instead, they map the frontier for future research and industry development, especially as multimodal and 3D AI models evolve.
A First Step Toward AI-Assisted Creative Exploration
The prototype does not replace storyboarding, nor is it meant to. Instead, it gives creators something invaluable: speed during the exploratory phase.
Teams can iterate more, try more ideas, and reach alignment earlier.
The human artist remains central.
AI simply clears the path.
Breda’s contribution to the AICI Innovation Alliance shows how thoughtful, responsible and creatively aligned AI development can support Europe’s creative industries without compromising artistic integrity.
It is a promising example of how AI can be woven into creative workflows in ways that respect craft, enhance collaboration and empower new generations of storytellers.