Creative Vocabulary

MovieLabs Creative Vocabulary

Executive Summary

MovieLabs Creative Vocabulary is a structured documentation of Film and Television terminology that for decades has been in use to describe creative storytelling choices across production workflows.

With input from subject matter experts, creative professionals, and AI model builders, this vocabulary captures the language that creative professionals have been using to propose, communicate, and implement ideas within audio-visual works. As generative AI becomes increasingly integrated into media production, having a common and consistent vocabulary helps preserve and harness the established language of filmmaking so it can be used effectively when interacting with AI systems.

For AI model providers, the Creative Vocabulary provides a practical framework for improving how generative models understand creative concepts. It can help identify gaps in training datasets where certain filmmaking concepts are underrepresented, guide the re-labeling of data where terminology has been inconsistently applied, and support prompt interpretation or enhancement layers that recognize synonymous terms without requiring model retraining.

The vocabulary also enables consistent tagging of media assets, improving the precision and recall of search across large media libraries. This benefits productions, studios, and software vendors building semantic search and AI tagging capabilities, asset management and production tools by enabling more reliable discovery of creative elements and media assets across workflows.

By documenting and defining a consistent, structured vocabulary that supports artists, AI systems, and production tools, MovieLabs Creative Vocabulary aims to improve creative communication, interoperability across production systems, and the controllability of generative AI in media creation.

Introduction

MovieLabs Creative Vocabulary is a structured documentation of Film and Television creative terminology used across production workflows. It defines terms that creative professionals use to propose, convey, and implement storytelling choices, beginning with cinematography and expanding in future versions to other areas of production.

Film and television production has long relied on a shared language to communicate creative intent. Directors, cinematographers, editors, sound designers, and other artists routinely use terms such as Close Up, Tracking Shot, Eye Level, or J-cut to describe how a scene should be captured or presented. As generative AI systems increasingly become part of production workflows, it is important that this established language remains usable when interacting with those systems. MovieLabs Creative Vocabulary helps ensure that the language of filmmaking continues to function as a practical way to guide both human collaboration and AI-assisted content creation.

Unlike traditional film craft manuals that focus primarily on explaining technical techniques, MovieLabs Creative Vocabulary emphasizes the creative intent behind those techniques and concepts. This emphasis is particularly important when communicating with generative systems, where understanding the intended creative outcome is often more useful than describing the underlying technical parameters.

The vocabulary is intentionally designed to be structured, concise, and accessible, making it useful not only for artists but also for machine learning practitioners and developers training and building generative systems and production tools.

The design of MovieLabs Creative Vocabulary follows several guiding principles. Terms that are documented and defined are those that are used to drive creative intent in productions, meaning they represent meaningful storytelling or visual or sound design decisions rather than purely technical specifications. For example, when distinguishing between Jib and Crane camera movement, the differences in physical equipment and setup are less emphasized, while the creative impact of constrained camera motion versus epic or expansive reveals is highlighted because such equipment differences are not meaningful in a generative AI world.

Terms are also instructive and precise, avoiding vague aesthetic labels in favor of terminology that clearly communicates the creative choice. The vocabulary also prioritizes broadly recognized production terminology, avoiding regional jargon or vendor-specific language whenever possible. Alternative names are included when they are commonly used, while the primary name reflects the most widely accepted term.

MovieLabs Creative Vocabulary has been developed in the MovieLabs Industry Forum through collaboration among Forum members, industry professionals, subject matter experts, generative AI technologists, and tool developers. Their combined expertise helps ensure that the terminology reflects both practical production usage and the emerging needs of AI-enabled workflows.

Use Cases

Creative Vocabulary supports two primary uses.

Guiding Generative AI with Established Film Language

Creative Vocabulary enables artists to continue using established filmmaking language when controlling or guiding generative AI models for image, video, audio, or 3D object creation. As artists increasingly interact with AI tools through prompts or structured inputs, the vocabulary provides a consistent set of creative concepts that can be understood by both artists and AI systems. This allows creative professionals to express intent using the same terminology they already use in traditional production workflows.

Improving Media Discovery Through Consistent Asset Tagging

Creative Vocabulary also provides a consistent reference for tagging and describing media assets. By applying standardized terminology to media, productions can search, discover, and organize assets more effectively. Consistent tagging improves both the precision and recall of asset searches, making it easier to locate relevant materials within large media libraries and archives. This consistency is particularly valuable when artists work across multiple productions, studios, and asset management systems, allowing production users to find assets using uniform terminology regardless of the specific project or tool involved.

Target Audience

By targeting the above two uses, MovieLabs Creative Vocabulary serves multiple audiences.

AI Model Providers and Model Aggregators

AI model providers can use the vocabulary to improve how generative models interpret creative concepts. The vocabulary can support several aspects of model development and deployment:

  • Training data acquisition: Identifying gaps where existing datasets do not sufficiently represent certain creative concepts defined in the vocabulary, signaling product development teams when new training datasets need to be acquired. For example, the vocabulary may expose to an AI Model Provider that Camera Angle concepts are not supported by their model because there are not enough examples that represent those concepts in their training dataset.
  • Data labeling and correction: Improving labeling consistency when datasets contain ambiguous or incorrect terminology, which can lead models to produce unintended results, again signaling product development teams when training datasets need to be re-labeled. For example, although there are enough examples in the training dataset, perhaps the Extreme Close Up and Close Up shots are not labelled properly for the AI model to differentiate between those concepts.
  • Prompt interpretation and enhancement: Supporting synonyms or alternative terms in prompting layers or prompt-enhancement systems, allowing models to recognize different ways users may refer to the same concept without necessarily requiring retraining or realignment of the model. Model aggregators and platforms that support multiple generative models can leverage MovieLabs Creative Vocabulary to normalize and translate user intent across models. These systems can accept prompts expressed using standardized creative terminology and dynamically rewrite or expand them into model-specific prompts based on each model’s capabilities and supported vocabulary. This allows users to express creative intent once, using familiar language, while enabling the aggregator to adapt that intent for different models without requiring users to learn model-specific prompting nuances.

These approaches help improve a model’s ability to understand and respond to the creative language commonly used in film and television production.

Software Vendors

Software vendors building production tools can incorporate the vocabulary into their systems in multiple ways.

  • AI-powered tagging and semantic search systems can leverage MovieLabs Creative Vocabulary. Vendors and platform providers building automated media analysis, indexing, and agentic workflows can use the vocabulary for labeling shots, frames, and sequences based on detected creative attributes. Aligning AI-generated metadata with the vocabulary enables consistent, production-relevant tagging at scale, improving both the precision and recall of semantic search. This allows users to discover content based on creative intent, such as shot size and camera movement rather than relying on inconsistent free-text tags or low-level visual features.
  • Asset management systems can provide interfaces that allow media assets to be tagged using the standardized terminology defined in the vocabulary. Consistent tagging enables improved search and discovery of assets across productions and media libraries.
  • Pre-production and planning tools can also benefit from the vocabulary. For example, applications used for storyboarding, shot lists, or production planning can incorporate vocabulary terms directly into their interfaces, enabling filmmakers to describe creative intent using consistent terminology throughout the planning and production process.

Creative Professionals

Creative professionals benefit from a structured and easy-to-read reference that reflects the language already used across film and television production. The vocabulary also provides an accessible introductory resource for emerging creators who may be new to filmmaking but are increasingly using generative tools to produce visual content.

Organization

The first version of Creative Vocabulary focuses on cinematography, covering the Lens, Shot Size, Camera Movement, and Camera Angle categories. These concepts form a foundational part of visual storytelling and are commonly referenced throughout production workflows.

Future versions are expected to expand the vocabulary to include additional categories such as Composition, Framing, and Depth of Field, as well as additional creative domains beyond cinematography.

The vocabulary is organized into creative areas, each containing categories, which in turn contain individual terms. Each term captures structured information including a definition, the typical creative intent behind using the term, alternative names or synonyms, broader concept to indicate that the concept behind the term is a special case of another broader concept, notes, and reference examples. This structure allows the vocabulary to function both as a reference for artists and as a guide for AI building teams.

Note: MovieLabs is still working on reference examples. We expect future versions to include reference examples.