ALL YOU NEED TO KNOW ABOUT
STORYTELLING, CONTENT & SYNTHETIC MEDIA
Every term here is defined the way a producer needs it rather than the way a paper defines it: what it is, what it costs you when it goes wrong, and what to do about that. Each one has its own page with the questions people actually type.
- Terms
- 162
- Topics
- 9
- Each entry
- Definition, working notes, questions
- Gate
- None
Acceptance Rate
The proportion of generated outputs that survive review and ship, measured per shot type.
Act Break
The point at which the story’s direction changes decisively enough that what follows operates under different rules from what preceded it.
Agentic Workflow
A process where a model plans and executes several steps against real tools, with a person deciding at named checkpoints instead of approving every keystroke.
AI Act Article 50
The EU transparency provision, enforceable since 2 August 2026, requiring deployers of deepfake and synthetic-media systems to disclose that content is artificially generated.
AI Slop
Generated content produced at volume without editorial judgement, recognisable by its smoothness, its symmetry and its complete absence of a point of view.
Answer Engine Optimisation
Structuring content so an AI assistant can quote it accurately, which mostly means answering the question in the first fifty words.
Antagonist
Whatever stands between the protagonist and what they want: a person, an institution, a condition, or the protagonist themselves.
Aspect Ratio Matrix
The full grid of placements, ratios and durations a campaign has to deliver, decided at brief stage because it changes how frames are composed.
Attention
The mechanism that lets a model weigh every part of its input against every other part, and the reason long sequences cost disproportionately more.
Autoregressive Video Model
A video model that generates forward in time, each segment conditioned on what it already produced, with no single pass denoising the whole clip.
C2PA / Content Credentials
An open standard that attaches a signed, tamper-evident record of how a piece of media was made and edited, travelling with the file itself.
Call to Action
The single thing you want the audience to do next, stated once, in the place where they are most persuaded, wherever that falls.
Camera Control
Directing a generative video model with the vocabulary of a camera department (dolly, truck, crane, rack focus) instead of describing what the shot should feel like.
Camera Path
The defined route a virtual camera takes through a shot, specified as a movement.
CFG Scale
A dial controlling how literally a diffusion model obeys the prompt, trading obedience against image quality.
Character Arc
The change a character undergoes across a story. What they believed at the start, what the plot costs them, and what they believe at the end.
Character Consistency
Keeping the same face and body recognisably identical, wardrobe included, across every shot and every session.
Checkpoint
A saved snapshot of a model’s weights at a point in training, and the thing you pin when you say a pipeline is reproducible.
Chekhov’s Gun
The principle that a conspicuous element introduced in a story must be used, and that anything which is not used should be removed.
Cliffhanger
Ending a unit at a moment of unresolved tension, so that stopping costs the viewer something.
Climax
The scene in which the dramatic question is answered, in which the protagonist makes the decision the whole story has been arranging, and after which nothing further can be argued.
Cold Open
A scene placed before any titles or branding, ahead of the setup, whose job is to establish a reason to stay before the piece formally begins.
Colour Management
Keeping colour consistent from generation through grade to delivery, and the reason a set of individually good clips can look like nine different films.
Conflict
The friction between what a character is trying to do and what prevents it, present at every scale from a line of dialogue to a whole story.
Connector
The link between a model and a piece of software a team already runs, and usually the difference between a system that gets adopted and one that does not.
Consent File
The folder of signed permissions and licences that has to exist before the first render.
Content Atomisation
Breaking one long piece into a set of claims and rebuilding each as a native asset for a different format.
Context Engineering
Deciding what a model sees for a given task, which is where the outcome is decided and the wording is not.
Context Window
The maximum amount of input a model can consider at once, measured in tokens, and a hard ceiling on how much brief it can hold.
Controlling Idea
The argument a finished story makes, stated as a claim about how the world works.
ControlNet
A conditioning layer that constrains a diffusion model with structural input (a pose skeleton, a depth map, an edge trace) so composition survives a change of style.
Cost Per Accepted Asset
The only generation metric that means anything: total spend divided by the number of assets that shipped.
Credit Ceiling
An agreed maximum spend per asset, encoded in the pipeline so a run halts and asks for a decision instead of quietly spending past it.
Curiosity Gap
The discomfort produced by knowing that there is something specific you do not know, and the mechanism most hooks are built on.
Cutdown
A shorter version of a finished piece cut for a different placement, planned into the shoot and never cropped out of it afterwards.
Data Minimisation
Putting only the data a task needs into a model, and never keys, client data or unreleased assets.
Deepfake
Synthetic media that depicts a real, identifiable person doing or saying something they did not, whether or not the intent is malicious.
Deliverable Specification
The exact list of files a client receives, with formats, resolutions, colour space, naming and who owns what afterwards.
Denoising Steps
How many passes a diffusion model takes to walk an image from noise to a finished frame.
Denoising Strength
How far an image-to-image generation is allowed to move from its input, from a light retouch to a complete reimagining.
Denouement
What comes after the climax: the short passage in which the consequences settle and the audience is shown the new normal.
Depth Map
A greyscale image encoding how far each pixel is from the camera, used to hold a composition steady while everything else changes.
Derivative Training
The specific permission to train a model on supplied material so it can generate new output resembling the subject, which pre-2024 releases almost never grant.
Diffusion Model
The architecture behind most image and video generators: it learns to reverse a noising process, and generates by denoising random static into a frame that matches your conditioning.
Diffusion Transformer
A diffusion model whose denoising network is a transformer in place of a convolutional U-Net, which is what let video models scale.
Disclosure
Telling the audience that what they are looking at was made or materially altered by a generative model, in the place they will see it.
Distillation
Training a smaller or faster model to reproduce the behaviour of a larger one, which is how most fast generation modes are built.
Dramatic Irony
The condition in which the audience knows something a character does not, which converts an ordinary scene into a tense one without changing a word of it.
Dramatic Question
The single yes-or-no question a story raises early and answers at the climax, which is what holds an audience through the middle.
Drift
The slow, cumulative change in a subject across a set of generations (a face, a product, a colour) where no single frame looks wrong but the run does.
Embedding
A list of numbers representing a piece of text, an image or a clip, positioned so that similar things sit close together.
Entity Graph
The structured picture a search or answer engine builds of who you are and how you connect to places, people, services and topics.
Eval
A fixed set of test cases with known good answers, run against a workflow to catch it getting worse.
Exposition
The information an audience needs in order to follow a story, and the part of writing where most pieces lose people.
Fine-Tuning
Continuing a model’s training on a narrow dataset so it reliably produces one style, one subject or one house voice.
First-Last Frame
Handing a video model a start image and an end image and asking it to generate the motion between them.
Flow Matching
A training approach that learns a straighter path from noise to image, letting a model produce good output in fewer steps.
Foreshadowing
Placing information early that will matter later, so that when it does the audience feels the ending was earned.
Foundation Model
A large general-purpose model trained on broad data, intended to be adapted to specific tasks before anybody uses it.
Freytag’s Pyramid
A five-part model of dramatic shape (exposition, rising action, climax, falling action, catastrophe or resolution), drawn as a triangle with the climax at the apex.
Gate
A named, mechanical test a frame must pass before anyone sees it, with a stated action on failure.
Generative Engine Optimisation
Making a brand legible to generative systems as an entity: what it is, what it does, where it operates and what it can be cited for.
Guardrail
A hard limit on what a workflow can do, enforced by the system itself.
Hallucination
Confident invention: a model producing a certification mark, a statistic, a limb or a citation that does not exist.
Hero Frame
The single image a campaign is built around and every other asset is judged against for tone and light, and for the colour it has to sit beside.
Hook
The opening moment of a piece, whose only job is to buy the next three seconds.
Human in the Loop
A named person with authority to stop the run, placed at the points in a workflow where a wrong decision is expensive.
Identity Lock
Our pipeline for synthetic presenters: train the face once from a stills sheet, then run every subsequent variant from that trained identity.
Image-to-Video
Generating a moving shot from a still you have already approved, instead of from text alone.
In Medias Res
Beginning a story part-way through the action and supplying the context afterwards, so the audience is oriented by consequence.
Inciting Incident
The event that upsets the balance of a character’s ordinary life and makes the rest of the story necessary.
Inference
Running a trained model to produce an output, as distinct from training it, and the part you are billed for per use.
Inpainting
Regenerating a masked region of an existing image while the rest of the frame stays untouched.
Latent Space
The compressed mathematical space a generative model works in, where similar concepts sit near each other.
Likeness Rights
A person’s control over the commercial use of their appearance, which generative production engages on almost every job.
Lip Sync
Matching a generated or existing face’s mouth movement to an audio track, and the point at which a synthetic presenter becomes convincing or does not.
Logline
One sentence naming the protagonist, their goal, the opposition and what is at risk. The test of whether a story exists yet.
LoRA
A small trained adapter that teaches a large model one style, character or object without retraining the model itself.
MacGuffin
An object or goal that motivates the characters and whose actual nature is irrelevant to the audience.
Master Plate
The single locked reference frame of a subject that every later generation is built from, so the subject cannot wander across a set.
MCP
An open protocol for connecting a model to external tools and data through a standard interface, so a capability written once works across clients.
Midpoint
The turn at the centre of a story where the protagonist stops reacting and starts acting, usually triggered by information that reframes everything before it.
Model Licence
The terms governing what you may do with a model’s output, which differ per model, per version and sometimes per plan.
Montage
A sequence of shots that compresses time or accumulates meaning through juxtaposition, with no continuous action.
Motion Brush
A control that lets you paint which regions of a still image should move and in which direction, instead of describing motion in words.
Multimodal Model
A model that takes and produces more than one kind of input (text, image, audio, video) inside a single system.
Naming Convention
An agreed, machine-applied pattern for naming files, and the single highest-return thing to automate in a creative team.
Narrative Arc
The overall shape a story takes across its length: where tension rises and where it releases, plotted against the points the audience is asked to feel something.
Narrative Transportation
The state in which an audience is absorbed enough in a story that they stop evaluating its claims, which is why stories persuade where arguments do not.
Negative Prompt
A list of things to steer away from, applied as a counterweight to the main prompt.
Open Loop
An unanswered question deliberately left running so the audience carries a reason to keep watching into the next section.
Operator Stack
Our pipeline for the unglamorous layer: skills, code, connectors, naming, versioning and the ledger that makes a creative studio explainable to a finance team.
Opt-Out
A machine-readable signal that a work should not be used for model training, and a claim you should be able to honour on work you publish.
Orchestration
The layer that decides what runs when and what happens on failure, plus the points where a person is asked to decide.
Outpainting
Extending an image beyond its original edges, generating new content that matches what is already there.
Pacing
The rate at which new information, change or tension arrives, which an audience experiences as speed regardless of how fast anything on screen is moving.
Pattern Interrupt
A deliberate break in rhythm or framing, placed where attention is predicted to drop.
Payoff
The moment a promise made earlier is kept, and the source of most of the satisfaction an audience attributes to a piece.
Phantom Set
Our pipeline for any subject a customer can hold up against the picture: lock a master plate, specify the set once, then generate every angle from the plate.
Point of View
Whose knowledge and experience the audience is restricted to, which determines what can be withheld and therefore what suspense is available.
Pose Estimation
Extracting a figure’s joint positions as a stick-figure skeleton, then conditioning generation on it so the body does what you asked.
Premise
The situation a story starts from, stated as a condition with an inherent problem inside it.
Previsualisation
Rough visual planning of shots before production, now cheap enough with generative tooling that skipping it takes some explaining.
Product Cinematic
A short, highly controlled moving piece whose subject is a single product, built from a locked plate.
Prompt Engineering
Writing model instructions with the precision of a brief: what, in what order, with what constraints and what it must not do.
Prompt Weighting
Syntax for telling a model which words in a prompt carry the weight, so nobody has to repeat them and hope.
Protagonist
The character whose decisions drive the story and whose question the ending answers. Not necessarily the most sympathetic one, or the one with the most screen time.
Provenance
The recorded history of how a piece of media was made and edited, travelling with the file so a platform or a viewer can check it.
Proxy Render
A fast, low-resolution generation used to test composition and motion before committing budget to a full-quality pass.
RAG
Fetching relevant source material at run time and giving it to the model, so the answer comes from documents and not from memory.
Reference Image
A picture supplied alongside the prompt to steer composition, style, colour or subject.
Reference Strength
How heavily a model is told to weigh a supplied reference image against the text prompt when the two disagree.
Regional Prompting
Applying different prompts to different areas of one frame, so elements stop bleeding into each other.
Register
A deliberate visual dialect: a decision about lighting model, edge quality and how much the world is allowed to be wrong.
Retention Curve
The graph of how many viewers are still watching at each second, and the only editorial feedback in video that is not an opinion.
Reversal
A moment at which the situation inverts: an advantage becomes a liability, a win becomes a loss, an ally becomes an obstacle.
Rotoscoping
Isolating a moving subject from its background frame by frame, still the most reliable way to protect a real element inside a generated one.
Run Log
The automatic record of what a pipeline did, what it cost and what happened at each gate, which is what makes a system auditable.
Sampler
The algorithm that decides how a diffusion model steps from noise to image, trading speed against fidelity and consistency.
Scene Objective
What a character is actively trying to get from the other people in a scene, stated as something they could succeed or fail at within it.
Scroll Stopper
The visual property of a piece that arrests a feed scroll before any of its content has been understood.
Seed
The number that fixes a model’s starting randomness, making a generation reproducible.
Segmentation Mask
A map marking which pixels belong to which object, used to change one element of a frame without disturbing the rest.
Set Specification
A short written document fixing the light direction, colour temperature, surfaces and lens language for a whole set, written before anything is generated.
Setup and Payoff
A paired structure in which something established early returns later with new weight, and the primary source of the feeling that an ending was earned.
Shot List
The ordered list of shots a piece needs, with framing, movement and intent stated per shot, written before anything is generated.
Show, Don’t Tell
The instruction to give an audience evidence they can interpret instead of a conclusion they must accept.
Silhouette Overlay
A control gate that lays a finished render over its master plate at around forty per cent opacity to check the product outline has not drifted.
Skill
A self-contained folder of instructions and resources a model loads on demand for a particular kind of job, instead of one enormous instruction it always carries.
Stakes
What the protagonist stands to lose, stated specifically enough that the audience can picture the loss.
Story Beat
The smallest unit of change in a story: one exchange, action or realisation after which something is different from how it was before.
Story Circle
An eight-step compression of the journey structure drawn as a circle: a character in comfort wants something, enters an unfamiliar situation, adapts, gets what they wanted, pays for it, and returns changed.
Story Spine
A fill-in-the-blank sentence sequence (once upon a time, every day, until one day, because of that, until finally) that produces a causally connected story.
Style Transfer
Applying the visual character of one image or corpus to the content of another.
Subplot
A secondary storyline that runs alongside the main one and exists to say something the main plot cannot say about itself.
Substantiation
The evidence held to support a claim made in an advert, required regardless of how the advert was produced.
Subtext
The meaning running underneath what is literally said or shown, produced by the gap between what a character wants and what they are willing to say.
Synthetic Endorsement
A recommendation delivered by a person who does not exist, or by a real person’s likeness saying something they never said.
Synthetic Media
Any image, video, audio or text produced or materially altered by a generative model.
Synthetic UGC
Creator-style content (presenter to camera, phone-look, room lighting) produced with a trained identity and no booked creator.
Temporal Coherence
Whether a generated clip holds together across time: the same face, the same coat, the same number of fingers, from first frame to last.
Text-to-Video
Generating a moving shot directly from a written description, with no source image.
The Hero’s Journey
A cyclical story pattern in which a protagonist leaves the ordinary world, is tested in an unfamiliar one, and returns changed and carrying something.
Three-Act Structure
The most common shape for a dramatic story: a setup that establishes a want, a middle in which pursuing it makes things worse, and an end in which the question is answered.
Token
The unit a language model reads and writes, roughly a short word or a fragment of one. It is also the unit most models are billed by.
Tone
The attitude a piece takes towards its own material, which the audience reads within seconds and uses to calibrate everything after.
Tool Calling
A model requesting that a defined function be run and reading the result, which is the mechanism that turns a chat into a workflow.
Topical Authority
Being demonstrably thorough on a subject, across several pages, measured by coverage and depth.
Trained Identity
A face trained once from a sheet of stills, which then carries across sessions, models and formats without re-uploading a reference.
Turning Point
Any moment that redirects a story’s trajectory, of which act breaks and the midpoint are the largest examples.
Value Shift
The change in what something is worth between the start and the end of a scene, and the most reliable single test of whether a scene is doing anything.
Variational Autoencoder
The component that compresses an image into the compact latent space a diffusion model works in, and decodes it back to pixels afterwards.
Voice Cloning
Building a synthetic voice from recordings of a real one, either quickly from a short sample or properly from a long one.
Want vs Need
The distinction between what a character is consciously pursuing and what the story says they require, which is usually incompatible with the first.
Watermarking
Embedding a machine-detectable signal in generated media so it can later be identified as synthetic.
World-Building
Constructing the rules, texture and history of a story’s setting to the depth the story uses, and no further.
A term we should have defined and didn't?
Send it over. If it belongs in the working vocabulary of anyone shipping synthetic media, it belongs on this page.