CHAPTER 1 — GLOSSARY OF TERMS

ACTION

A prompt element describing what a subject does during a generation, such as walking, turning, sitting, winking, or removing a coat. In video generation, an Action normally consumes part of the available duration.

ALIGNMENT ERROR

The amount by which generated objects fail to maintain a specified alignment relationship. Alignment Error is one possible measurement of Relationship Fidelity.

ANCHOR

A fixed reference image supplied to a generation process to constrain a required visual state. An anchor may establish identity, composition, environment, geometry, subject position, or a required temporal starting or ending state.

ANCHORED GENERATION

Generation using fixed beginning and ending image states, with the renderer generating the motion connecting them. Anchored Generation is a temporal interpolation problem and is distinct from Native Generation and Compositing Generation.

Anchored Generation: fixed beginning and ending image states constrain the renderer-generated motion between them.

ANIMATE FRAMES

The RML workflow used in this research for anchored video generation from two or more image inputs. Two-keyframe Animate Frames has been the principal tested implementation of FLF2V. RML transaction records have identified Animate Frames generations under the billing/model name Veo 3.1, showing that the visible RML tool name and underlying billing identity may differ.

APPARENT SCALE

The size that a subject or object occupies within the two-dimensional image frame. Apparent scale is distinct from the object's real-world physical size and changes according to distance, focal geometry, and perspective.

ARTICULATION

A prompt element describing how an Action is performed rather than what the Action is. Examples include slowly, subtly, forcefully, with one hand, or one finger at a time.

ASPECT RATIO

The proportional relationship between the width and height of an image or video frame, expressed as a ratio such as 16:9.

ASSET

Any still image, reference image, generated image, video clip, extracted frame, audio file, or other media object used as an input, intermediate element, or output in the production workflow.

BACKGROUND PRESERVATION FIDELITY

The degree to which a Compositing Generation leaves the already-fixed background unchanged while adding or modifying specified content. It applies specifically to Compositing Generation because Native Generation has no pre-existing background to preserve.

BEAT

A discrete production unit representing one intended visual action, moment, still, or clip. The project file system organizes assets primarily by Scene and Beat rather than by date or tool because a single Beat may require multiple tools and working sessions.

BEHAVIOR

A failure-classification term describing whether a generated pose or action is consistent with the known real-world behavior of the subject.

BEHAVIORAL PARSER

The provisional failure-analysis category concerned with whether a renderer understands why a subject behaves or is posed in a particular way rather than merely recognizing that the pose is visually possible.

BEST-OF-N

A production method in which multiple generations are produced under the same conditions and the best acceptable result is selected. Best-of-N is legitimate for production, but benchmarking must retain the complete distribution of attempts rather than reporting only the keeper.

BEST-OF-N GAS

The highest Geometric Accuracy Score obtained among N attempts made under identical or equivalent test conditions. It is a proposed measure of Iteration Efficiency rather than a measure of single-generation reliability.

CAMERA

A prompt category describing movement or behavior of the image frame itself, independent of subject motion. Examples include pan, push in, orbit, tilt, or static/locked camera.

CAMERA HEIGHT

The vertical position of the camera above the defined Ground Plane. Camera Height is an extrinsic Camera Model parameter used in perspective calculations.

CAMERA MODEL

A defined and verified projection geometry describing how a camera is positioned, oriented, and optically configured for a shot. A Camera Model establishes the conditions required to calculate where subjects or objects should appear within the two-dimensional frame. Depending on the model and workflow, its parameters may include Camera Height, Pitch, Roll, Focal Length, Aspect Ratio, Horizon %, frame dimensions, and subject distance. Once validated, a Camera Model is intended to function as a reusable production specification rather than a newly invented description for each generation.

CAMERA MODEL CALCULATOR

The calculation tool developed to derive Camera Model geometry rather than estimate it visually. It calculates relationships including horizon placement and apparent subject size from Camera Model and subject parameters.

CAMERA MODEL INSTANTIATION

The process by which a renderer attempts to convert a Camera Model specification into an actual generated image having the intended camera geometry.

CAMERA MODEL LIBRARY

The collection of defined Camera Models, identified as CM-01, CM-02, and so forth, intended to provide reusable and individually verified camera configurations for production.

CAMERA MODEL PARAMETERS

The variables defining a Camera Model. Current research includes Camera Height, Pitch, Roll, Lens/Focal Length, Aspect Ratio, Horizon %, frame dimensions, and relevant subject-position information.

Camera Model Parameters: explicit geometric variables used to define and measure camera configuration.

CANVAS

The complete pixel dimensions of the delivered image frame. Canvas dimensions must be verified from the actual generated asset rather than assumed from prompt text or a tool's nominal resolution label.

CARDINALITY

Whether the renderer produces the correct number of specified entities or parts. Cardinality is independent of their size, position, or geometric accuracy and can generally be evaluated by counting.

CENTERLINE

The horizontal midpoint or other defined central reference line of the image frame. Centerline is used as an X-position reference when calculating subject or object placement.

CHAINED FLF2V

A workflow in which multiple FLF2V generations are connected sequentially. When the clips are intended to form one continuous moment, the actual final frame of the preceding clip should be used as the next starting frame rather than independently regenerated.

CINEMATIC INSTANTIATION

See Language Instantiation.

Camera Model. Individual Camera Models are numbered CM-01, CM-02, and so forth.

CMIT

Camera Model Instantiation Test. A controlled test determining whether and how accurately a renderer can instantiate the geometry specified by a Camera Model.

CMIT-C

Camera Model Instantiation Test — Compliance. The quantitative stage of Camera Model testing that asks how accurately a renderer achieved the requested magnitude after successful directional instantiation has been established.

CMIT-I

Camera Model Instantiation Test — Instantiation. The stage asking whether the renderer responded to the requested camera change at all and whether the response occurred in the correct direction.

CMIT-P

Camera Model Instantiation Test — Pitch. The test series developed to determine whether a renderer derives correct horizon placement from specified Pitch and whether explicit Horizon % can reliably control or override the result.

COMPILER

The translation layer between the Director's tool-independent production specification and the instructions required by a particular renderer. The research increasingly supports renderer-specific rather than universal compilation.

COMPOSITE FIDELITY

The complete evaluation checklist applied to a Compositing Generation. Composite Fidelity is not a separate score; it brings together applicable measures such as Constraint Fidelity, Background Preservation Fidelity, Semantic Fidelity, and Relationship Fidelity without collapsing them into a single metric.

COMPOSITING GENERATION

Generation in which new content is added to an already-fixed reference image. The existing image therefore acts as a constraint that the renderer is asked to preserve while introducing new content. It is distinct from Native Generation and Anchored Generation.

COMPOSITION BIAS

A renderer's tendency to favor a visually or semantically dominant portion of an existing reference image when positioning newly generated content, even when another placement has been requested.

COMPLIANCE

The degree to which a generated result actually follows the Director's specified instruction. A visually attractive or usable result may nevertheless be non-compliant.

CONSTRAINT FIDELITY

The degree to which a renderer honors explicit quantitative specifications such as apparent pixel height, Feet-Y position, resolution, Aspect Ratio, or other measurable targets. GAS is the principal metric developed for this evaluation axis.

CONSTRAINT SOLVER

The hypothesized capability of a generative renderer to treat an explicit specification, particularly a numerical specification, as a mandatory target rather than as an approximate semantic cue. The research has found this capability substantially less reliable than general semantic understanding.

CONVERGENCE

The event in which a moving subject reaches a specified spatial destination, scale, or target relationship.

The measured horizontal center coordinate of a subject or object. CX is used to compare actual X-position with a calculated target.

DETERMINISTIC PIXEL OPERATION

An image-editing operation that directly alters specified pixels while leaving all other pixels exactly unchanged. It is distinct from Generative Editing, which may regenerate supposedly untouched portions of the image.

DIRECTOR

The human creative authority who determines the intended visual result and decides which generated outputs are acceptable. In this methodology the Director determines the load-bearing creative choices; the renderer is expected to execute them.

DIRECTOR'S DECISION

A deliberate decision to retain a generated result despite its failure to comply fully with the original instruction because the Director determines that the result has independent creative value. Acceptance does not retroactively convert non-compliance into compliance.

DIRECTOR'S DECISION FRAMEWORK

The formal rule separating compliance from usefulness: non-compliance is recorded as failure; a failed output may nevertheless possess independent value; and using it is a conscious Director's Decision.

ENVIRONMENT

A prompt element describing the world surrounding the subject, including location, time of day, lighting context, setting, and other scene conditions that generally remain static unless otherwise instructed.

EXTRINSIC CAMERA PARAMETERS

Parameters describing where the camera is located and how it is aimed relative to the scene. These include Camera Height and camera orientation such as Pitch and Roll.

FAILURE TAXONOMY

The evolving classification system for distinguishing different kinds of renderer failure rather than treating all unwanted output as generic prompt failure. The current research includes parser-based categories and cross-cutting violated-constraint categories.

FEET-Y

The vertical image coordinate of a standing subject's ground-contact point at the feet. Feet-Y is used as a repeatable measure of perspective placement and Camera Model compliance.

Frame Coordinate Measures: Top-Y, Feet-Y, and CX provide repeatable two-dimensional subject-position measurements.

FIRST FRAME

The first image state of a video generation.

FIRST STILL

A fixed still image supplied as the required starting state of an Anchored Generation.

FIXED TECHNICAL CONSTANTS

Values required by Camera Model mathematics but not normally supplied to the renderer as prompt instructions. Examples include assumed Sensor Height and Sensor Width.

FLAT-EARTH HORIZON

The theoretical geometric horizon defined relative to an ideal flat, level Ground Plane. Mountains, terrain, buildings, trees, and other visible content do not change its location.

FLF

First/Last Frame. Shorthand for a workflow in which fixed first and last visual states constrain the generated motion.

FLF2V

First-Last-Frame-to-Video. Video generation using a fixed First Still and fixed Last Still, with the renderer generating the intervening motion.

FMFLF2V

First-Middle-Frame(s)-Last-Frame-to-Video. A control architecture using a fixed First Frame, one or more intermediate image states, and a fixed Last Frame as temporal anchors within a video generation. A user-facing First/Middle/Last keyframe architecture existed historically in Runway Gen-3 Alpha Turbo but is not currently available in the required explicit form among tested RML workflows. References are not equivalent to temporally positioned keyframes. The production reliability and internal implementation of FMFLF2V remain unvalidated by this research.

FOCAL LENGTH

An optical property of a camera lens, expressed in millimeters, that affects Field of View and apparent perspective. It is an intrinsic Camera Model parameter.

FOV

Field of View. The angular extent of the scene represented within the frame.

FRAME

A single two-dimensional image, whether generated independently as a still or existing as one temporal image within a video clip.

FRAME INSTABILITY

Variation in camera geometry or composition across repeated generations made from the same prompt and nominally identical settings.

GEOMETRIC ACCURACY SCORE

A 0–1 measure of how closely one generated spatial property matches its calculated target. GAS = 1 − (|Actual − Expected| ÷ Expected) GAS can be applied separately to apparent height, X-position, Feet-Y, horizon location, vanishing point, or another calculable property.

GENERATION

One execution of a generative tool producing an image, video, or other output from supplied assets and instructions.

GENERATIVE EDITING

Image modification performed by a generative model. Even when instructed to alter only one region, the model may regenerate texture, edges, shadows, or other supposedly unchanged portions of the image.

GENETICS

A provisional failure-classification term describing inconsistent attributes within a single generated entity, such as incompatible coloring or markings appearing on different parts of the same subject.

GEOMETRIC HORIZON

The horizon determined by projection geometry rather than by the visible boundary between land and sky.

Horizon Definitions: geometric horizon, ideal flat-ground horizon, and the visible land/sky boundary are distinct concepts.

GEOMETRIC STABILITY

The degree to which repeated generations preserve the same underlying camera or spatial geometry even though their actual pixel content may differ substantially.

GROUND PLANE

The idealized level surface upon which subjects are assumed to stand for purposes of Camera Model and perspective calculations.

HORIZON %

The vertical location of the Geometric Horizon expressed as a percentage of total frame height measured from the top of the image.

IMAGE-TO-VIDEO

Video generation using an existing still image as an input or starting reference. Image-to-Video may be open-ended with one reference image or anchored when additional required image states are supplied.

INSTRUCTION LEAKAGE

A failure in which structural or metadata text intended solely as an instruction, such as a Camera Model label, ID, or parameter, is rendered visibly in the output image.

INSTANTIATION

The act of converting an abstract specification into an actual generated result. In the Camera Model system, instantiation concerns whether the renderer establishes the requested camera geometry.

INSTANTIATION METHOD

The form in which Camera Model instructions are presented to a renderer. The currently defined methods are Parametric Instantiation and Language Instantiation.

INTRINSIC CAMERA PARAMETERS

Parameters belonging to the camera or lens itself rather than to its position in the scene. These include Focal Length or equivalent Field of View and the relevant frame or sensor dimensions.

ITERATION

One generation attempt within a sequence of attempts directed toward the same intended result.

ITERATION EFFICIENCY

A proposed measure of the practical effort required to obtain an acceptable result across repeated generations. It is distinct from GAS, which evaluates one generated result.

ITERATIONS-TO-THRESHOLD

The number of attempts required before a generation reaches a predefined acceptable performance threshold.

KEEPER

A generated asset selected for production use because it satisfies the Director's requirements or has been deliberately accepted through a Director's Decision.

KEYFRAME

A still image used to establish a required visual state at a defined point in a video-generation sequence.

LANDMARK

A clearly identifiable feature within an existing image used as a positional reference when instructing where another subject or object should appear.

LANDMARK STRENGTH

The degree to which a visual landmark provides an unambiguous positional reference to a renderer. Strong landmarks are distinctive and spatially precise; weak landmarks are ambiguous or visually diffuse.

LANGUAGE INSTANTIATION

Camera Model instantiation expressed through natural or cinematic language such as “low angle shot” or “camera looking upward,” rather than through numerical parameters.

LAST FRAME

The final frame of a generated video clip. When chaining clips into a continuous moment, the actual Last Frame should normally become the next clip's First Still.

LAST STILL

A fixed still image supplied as the required ending state of an Anchored Generation.

LATERAL OFFSET

The horizontal displacement of a subject or object from the camera or frame Centerline.

LENS

The optical specification controlling Focal Length and Field of View. Within Camera Models, lens is normally expressed as Focal Length in millimeters.

LOCOMOTION WITHOUT DISPLACEMENT

A failure mode in which a subject visibly performs a walking or locomotion cycle without making the requested spatial progress toward its destination. Motion occurs, but displacement does not.

MASTER

A verified reference asset selected as an authoritative source for subsequent production generations or controlled tests.

MASTER A / B / C

Hash-identical copies of one verified Master image used so separate compositing tests begin from provably identical background conditions.

MEASUREMENT PRECEDENCE

The methodological rule that direct measurement of the actual generated asset takes precedence over visual impression, memory, filenames, interface labels, or expected behavior.

MDwC

My Dinner with Claude, the screenplay and trailer project from which much of the methodology research originated.

MIDDLE FRAME

An intermediate image state positioned temporally between the First Frame and Last Frame of a video sequence.

The number of observations or generations in a test condition. For example, n=3 means three independent generations were performed under that condition.

NANOB

Project shorthand for Nano Banana. Because multiple Nano Banana versions are available, formal testing should identify the exact version when known.

NATIVE GENERATION

Generation in which the model constructs the entire scene in one pass without an existing reference image whose content must be preserved.

NATIVE STILL GENERATION

Native Generation producing a still image.

NATIVE VIDEO GENERATION

Video generation in which the renderer creates the moving scene without a required fixed ending state.

OCCLUSION

The visual blocking of one subject or object by another because one lies in front of the other from the camera's viewpoint.

OCCLUSION ACCURACY

A measure of whether requested front/behind and visual-blocking relationships are rendered correctly. It is one possible measure of Relationship Fidelity.

OPEN-ENDED GENERATION

Generation without a fixed required final state. In video, the renderer begins from text or a single starting reference and is free to determine much of the subsequent motion and ending.

OUTPUT RESOLUTION

The actual pixel width and height of the delivered generated asset. Output Resolution must be measured from the file and should not be assumed from interface labels such as 1K, 2K, or 4K.

OVERCONFIDENCE PROTOCOL

The methodological discipline requiring findings to be stated only as broadly as the evidence supports, with sample size, scope, outliers, untested conditions, and provisional status made explicit.

PARAMETRIC INSTANTIATION

Camera Model instantiation using explicit structured or numerical parameters such as Pitch, Horizon %, Camera Height, or Focal Length rather than natural cinematic language.

PDS

Prompt Difficulty Score. A pre-generation risk estimate of the likelihood that a renderer will fail one or more evaluation axes. The current framework adds one point for each identified difficulty factor. PDS predicts risk; it does not measure the output.

PDS-TO-ITERATION MAPPING

A proposed future predictive system relating Prompt Difficulty Score to expected Iterations-to-Threshold and likely failure modes. The present research does not yet contain enough data to establish the mapping reliably.

PHYSICAL PARSER

The provisional failure-analysis category governing internal physical consistency, including lighting, shadows, occlusion, and related scene physics.

PHYSICS

A violated-constraint category covering physically impossible or inconsistent lighting, shadows, occlusion, or related relationships.

PITCH

Vertical rotation of the camera's viewing direction relative to level. Pitch affects the position of the Geometric Horizon in a real projection system.

PIXEL DIFF

A direct numerical comparison between two images measuring differences in their pixel values.

PIXEL STABILITY

The degree to which actual pixel content remains identical or nearly identical across generations or transformations. Pixel Stability is distinct from Geometric Stability; substantially different pixels can coexist with highly consistent camera geometry.

PRINCIPAL POINT

The optical center of the projected frame. In the simplified level-camera model used in this research, the Geometric Horizon passes through the vertical center of the frame when Pitch is zero.

PROMPT

The human-readable instruction supplied to a generative system describing desired content, action, state, camera behavior, timing, geometry, or other requirements.

PROMPT TAXONOMY

The working grammar developed to classify prompt elements according to their function. Current categories include State, Environment, Action, Articulation, Camera, Sequence, Timestamp, and Stillness/Idle.

PROXY PARAMETER SUBSTITUTION

A failure mode in which the renderer responds to the qualitative intent of an instruction by changing a correlated but different parameter rather than the parameter actually specified. The originating example was a “low angle” instruction that substantially changed apparent proximity/FOV while leaving the measured Horizon % nearly unchanged.

PULSE REFERENCING

A proposed but not project-validated technique in which a reference image is reintroduced at intermediate moments during video generation to reinforce identity or state before difficult deformation events. It remains an unconfirmed technique rather than an established RML capability.

QUANTIFIER SCOPE

A failure class in which an instruction intended to apply to a specified subset of entities is instead applied to the whole class. An example is “one eye winks” resulting in both eyes winking.

RADIAL DISTANCE

The straight-line distance from the camera position to the subject position. The term was adopted to remove ambiguity from the phrase “X feet from camera.”

RATIO ERROR

The amount by which a generated proportional relationship differs from the specified proportional relationship between multiple subjects or objects.

REFERENCE IMAGE

An existing still supplied to a generative tool to constrain some aspect of the output, including environment, identity, geometry, composition, style, or state.

REFERENCE PLATE

A fixed Reference Image whose geometry and relevant content have been verified before use as the base for Compositing Generation or controlled testing.

REFERENCE PRESERVATION

The degree to which portions of a Reference Image intended to remain unchanged actually remain unchanged through an editing or generation process.

REFERENCE PRESERVATION TEST

A test determining whether an editing or generation method preserves specified regions of a Reference Image to the required standard, including pixel-exact preservation where necessary.

RELATIONSHIP FIDELITY

The degree to which a renderer preserves specified relationships between multiple subjects or objects, such as scale ratios, ordering, spacing, alignment, and occlusion, independently of whether their absolute values are correct.

RELATIVE POSITION ERROR

The deviation between the requested positional relationship among multiple objects and the relationship actually generated. It is one possible measurement of Relationship Fidelity.

RENDERER

The generative model or system that converts supplied assets and instructions into visual output. Within this methodology, the Renderer is distinguished from RML as the platform/wrapper and from the Director as the source of creative intent.

RENDERER RESPONSIVENESS

The degree to which a renderer produces a measurable change in response to an instruction. Responsiveness establishes that the instruction affected the generation; it does not establish that the amount or mechanism of the response was correct.

RENDERER-SPECIFIC COMPILER

A translation layer calibrated to the measured behavior of a particular renderer. It converts a common production or Camera Model specification into the form of instruction most likely to instantiate that result on that renderer.

RESOLUTION LABEL

A user-interface designation such as 1K, 2K, or 4K describing a nominal output tier. A Resolution Label must not be treated as equivalent to verified Output Resolution unless the actual generated file confirms it.

RML

Runway ML. In this methodology RML is treated as the platform or wrapper containing multiple models, tools, modes, asset views, and workflows rather than as a single generative model.

ROLL

Rotation of the camera around its forward viewing axis, producing a tilted horizon.

SEMANTIC FIDELITY

Whether a generated scene preserves the intended meaning of the instruction independently of its geometric accuracy. A formal measurement method remains less developed than those used for Spatial Geometry.

SEMANTIC PARSER

The provisional failure-analysis category governing whether the renderer correctly interprets prompt words and descriptive relationships.

SEMANTIC RENDERER

The hypothesized strong layer of a generative visual model that understands objects, scenes, and qualitative relationships well enough to produce plausible imagery even when exact numerical constraints are not obeyed. It is contrasted with the weaker Constraint Solver.

SEMANTICS

A violated-constraint category describing cases in which a word or relationship in an instruction is interpreted with a different meaning from the one intended.

SENSOR HEIGHT

The assumed physical vertical dimension of the camera sensor used in Camera Model calculations to convert Focal Length into vertical Field of View.

SENSOR WIDTH

The assumed physical horizontal dimension of the camera sensor used in Camera Model calculations.

SEQUENCE

A natural-language temporal structure specifying the order of actions through terms such as “then,” “after,” and “finally.” Sequence and Timestamp express temporal ordering at different levels of precision.

SEQUENTIAL PERSPECTIVE

Perspective behavior occurring continuously through motion rather than only at isolated static positions. It concerns whether apparent scale and spatial relationships change correctly as subjects move toward or away from the camera.

SESSION

An RML working context containing generations and associated history. The binder recommends descriptive Session names so prior work can be found reliably rather than depending solely on recent-history views.

Stick Figure. Project shorthand for simplified human figures used to isolate spatial placement, scale, perspective, and motion from unnecessary visual complexity.

Spatial Geometry. SG-### identifies numbered and reproducible Spatial Geometry benchmark scenes used in place of ad hoc test prompts.

SPATIAL GEOMETRY

The research area concerned with measurable placement, distance, apparent scale, perspective, ground contact, and relationships among subjects or objects within the image frame.

SPATIAL PARSER

The provisional failure-analysis category governing placement, distance, scale, and contact between figures or objects. It is currently the most quantitatively developed parser.

SPT

Sequential Perspective Test. The experimental framework for determining whether perspective-correct spatial relationships, particularly changing apparent scale, can be maintained through continuous motion.

STATE

A defined visual condition existing at a particular point in a production sequence. A State may include identity, wardrobe, pose, object condition, environment, composition, spatial relationships, or other visible attributes. In the Prompt Taxonomy, State refers more narrowly to fixed attributes that should remain constant unless otherwise instructed.

STATE PERSISTENCE

The ability of a renderer to maintain a required condition, identity characteristic, pose, object, or other specified state through motion or across time.

STILL

A single static image used as a finished visual asset, reference, anchor, or intermediate production state.

STILLNESS / IDLE

An explicit instruction assigning otherwise unused video duration to controlled stillness or idle behavior rather than allowing the renderer to invent additional Action.

TEXT-TO-VIDEO

Video generation initiated primarily from textual instruction rather than from a required image anchor.

TIMESTAMP

An explicit numeric time marker assigning actions or states to portions of a video's duration. Timestamping is the more precise counterpart to natural-language Sequence prompting.

TOOL

A specific generative or editing capability accessible within or alongside RML. Tools should not be assumed to share capabilities merely because they appear within the same interface.

TOOL SELECTION PROTOCOL

The rule requiring the exact tool or model to be identified before generation and selected according to the required technique rather than habit, familiarity, or launcher position.

TOPOLOGY

A failure category involving impossible or incorrect connections among independent entities or anatomical parts, such as two subjects becoming fused into one structure.

TOP-Y

The vertical image coordinate of the topmost defined point of a subject or object. Together with Feet-Y and apparent height, Top-Y is used to measure vertical placement.

TRAJECTORY

The spatial path followed by a moving subject or object through a video sequence.

TRAJECTORY CONVERGENCE

A motion test asking whether a moving subject follows its intended trajectory while changing perspective appropriately and ultimately reaches the specified destination or target relationship.

UNDERDETERMINED

A condition in which insufficient information has been supplied to produce one mathematically or spatially unique correct solution. A positioning instruction is underdetermined when one or more parameters required to calculate the intended position are absent.

VANISHING POINT

The point in a two-dimensional perspective image toward which parallel receding lines converge. Vanishing-point analysis is the preferred method for determining the true Geometric Horizon when terrain or other image content makes visual land/sky boundaries unreliable.

WORKFLOW

The ordered sequence of assets, generation modes, tools, transformations, verification steps, and handoffs used to produce an intended visual result.

X-POSITION

The horizontal coordinate of a subject or object within the two-dimensional frame.

Y-POSITION

The vertical coordinate of a subject or object within the two-dimensional frame. Specific Y-position measurements used in this methodology include Top-Y and Feet-Y.

END CHAPTER 1