Image / Graphic
Represent one managed still visual asset and its technical, descriptive, provenance and use evidence across referenced representations.
Bundle → Layer → Finding → Questions Filled
3 bundles · 4 layers · 5 findings · 11 questions
Asset and renditions Which image this is and in which forms it exists.
Master and derivatives
The master file and the renditions derived from it.
Master file
The original or highest-quality file with its format and dimensions.
- Which file is the master, and in which format and pixel dimensions or vector form?
- Which colour space and bit depth does the master use?
Derived rendition
A resized, cropped or converted version and its link to the master.
- From which master and by which transformation was this rendition made?
- Has the rendition been edited in ways that change what it shows?
Provenance and rights Where the image comes from and how it may be used.
Origin and authorship
Capture or creation, creator and edit history.
Capture record
When, where and with what the image was captured or created.
- Who created the image, and when and where was it captured or made?
- Is a signed provenance manifest attached, and does it verify?
- Was the image generated or substantially altered by an AI system?
Licence and consent
Rights, licence and consent of depicted people.
Usage terms
The licence, credit line and any model or property release.
- Under which licence may the image be used, and what credit is required?
- Are identifiable people depicted, and is there consent or a release for this use?
Description and access How the image is described for search and accessibility.
Descriptive metadata
Caption, keywords and text alternative.
Accessible description
A caption and text alternative that describe the content.
- Does the image have a caption and a text alternative for screen readers?
- Are keywords and subject terms drawn from a controlled vocabulary?
Classifiers Filled
- Family
- World Models
- Category
- Information and virtual systems
- Entry kind
- entity
- Navigation path
- NAV.INF.MED.IMG
- Domain
- INF.MED.IMG
- Industry
- Cross-industry
- Tags
- imagegraphicinf.med.img
What it is Filled
An image or graphic is a still visual asset, such as a photograph, illustration, diagram or icon, held as a raster or vector file together with its technical and provenance metadata. The class covers one still asset and its renditions; moving images, 3D scenes and the creative work as a legal object are described in their own models.
In scope
- Photographs, illustrations, diagrams and icons in raster, vector or mixed still form.
- Image-specific observations, derivation references, descriptions, use assessments and continuity evidence.
- Explicit still extractions from larger sources with source selectors and lineage.
Out of scope
- Creative-work and rights-contract lifecycle, publication operations and distribution execution.
- Audio/video tracks, animation behavior, document page ordering and 3D scene management.
- Rendering, verification, recognition, server operation, erasure and legal adjudication engines.
- Physical carrier or depicted-object measurements except referenced calibration evidence.
Why it exists Filled
Represent one managed still visual asset and its technical, descriptive, provenance and use evidence across referenced representations.
Distinguishing features Filled
- It is a single still visual asset with renditions, not a video, audio recording or 3D scene.
- Technical metadata such as dimensions, colour space and format is part of the asset record.
- Provenance and edit history matter because images are easily altered or generated.
- It differs from the creative work as a rights object, which may have many images as expressions.
What robots and AI may and may not do Filled
Must not
- Strip creator, licence or provenance metadata from an image.
- Present a generated or altered image as an unaltered photograph.
- Use an image beyond its licence or without required credit.
- Identify people in images by facial recognition without a lawful basis and consent where required.
Only with a human decision
- Publishing images of identifiable people without a recorded release.
- Approving edits that change what a news or evidential image shows.
- Licensing images to third parties.
May
- Read and write technical, descriptive and rights metadata in standard fields.
- Create renditions such as thumbnails while keeping the link to the master.
- Verify provenance manifests and report the result.
- Write text alternatives that describe image content.
Moral aspects Filled
- Images of people affect their privacy and dignity, especially of children and people in vulnerable situations.
- Manipulated or generated images can mislead the public and damage reputations.
- Creators depend on credit and licence terms being respected.
Who is affected
- People depicted in images
- Photographers, illustrators and other creators
- Audiences who rely on images as evidence
Owners Filled
Steward
The rights holder or the organization that commissioned or licensed the image, usually managed by a digital asset librarian.
Roles
- Image steward
- Resolve identifiers, scope and record revisions.
- Technical reviewer
- Assess bounded inspection, rendering dependencies, mappings and fitness.
- Rights and disclosure reviewer
- Assess permissions, releases, sensitive metadata and intended-use conditions.
- Accessibility reviewer
- Review placement-specific alternatives and presentation evidence.
- Preservation custodian
- Own retention and storage disposition execution and supply receipts.
- Record agent
- Populate evidence-qualified fields within authorized scope; escalate missing authority.
Master systems
- Digital asset management systems
- Stock and picture libraries
- Museum and archive collection systems
Links to other meta-models Filled
extends
- WM-MED-002 - Proposed specialization of the registry parent Media Asset / Rendition. Generic media identity, storage and rendition lifecycle remain external; only still-image parameters and linked evidence are local. Pin a reviewed version before runtime composition.
references
- WM-MED-001 - Optional creative-work reference; intellectual-work identity and rights lifecycle are not owned here.
- WM-MED-003 - Optional publication/edition reference for placement context; publication scheduling and distribution lifecycle stay external.
- WM-MED-004 - Optional source recording/frame reference. Time-based tracks and animation lifecycle remain outside this still asset; extraction gets an explicit revision and time/frame selector.
- WM-MED-006 - Optional source scene reference for rendered images; geometry and scene lifecycle are not reproduced here.
aligned
- https://iptc.org/std/photometadata/specification/IPTC-PhotoMetadata-2025.1.html - Conceptual metadata mapping only; exact field bindings, vocabularies and round trips need validation.
- https://iiif.io/api/image/3.0/ - Optional image-delivery interface; no server operation or compliance claim.
- https://www.w3.org/TR/prov-o/ - Conceptual attribution and derivation links; no complete ontology mapping asserted.
- https://spec.c2pa.org/specifications/specifications/2.2/specs/C2PA_Specification.html - Reference credential and verifier outcomes under a pinned version and trust policy; no implemented validator.
neighbor
- WM-MED-002 Media Asset / Rendition - Registry parent suggests specialization; no relation-ledger row exists. Generic asset identity and operational lifecycle remain mastered externally; the EXTEND link is proposed and unpinned.
- WM-MED-001 Creative Work - Work identity and legal status are not inferred from the existence of image bytes or creator metadata.
- WM-MED-003 Publication / Edition - Only intended-use and placement references are local; publication and distribution remain external.
- WM-MED-004 Audio / Video Recording - An animation or recording is outside this root; a selected still frame has explicit source and selector evidence.
- WM-MED-006 3D Asset / Scene - A rendered still can reference a scene without owning geometry, animation or scene lifecycle.
- Digital repository, rights and audit masters - Record supporting evidence and decisions without duplicating storage, enforcement or audit-trail execution.
parent
- WM-MED-002
What else AI and robots need to interact with it Filled
Identity and identifiers required Filled
- An image is identified by an asset identifier in its managing system, and often by a content hash of the master file.
- Published images may carry persistent identifiers such as DOIs and identifiers in embedded IPTC or XMP metadata.
Direct properties not applicable Not applicable
Not applicable
An image or graphic is a digital asset; its pixel dimensions, colour space and format are technical metadata recorded in the structure, not physical measurements.
Recognition optional Filled
- An image is recognized by its file format, dimensions and embedded metadata.
- Often confused with a frame from a video, with a document page scan or with the work it depicts.
Capabilities and actions required Filled
- Can be rendered in different sizes and formats from a master.
- Can carry embedded metadata and signed provenance across platforms.
- Can be searched by description, keywords and visual similarity.
Hazards and failure modes required Filled
- Misinformation through manipulated or out-of-context images.
- Privacy harm from publishing images of identifiable people.
- Copyright infringement when licence terms are lost or ignored.
Standards and interfaces required Filled
- Exif for capture metadata and IPTC Photo Metadata in XMP for descriptive and rights fields.
- C2PA content credentials for signed provenance.
- IIIF Image API for delivering images from collections.
Context of use required Filled
- Used in news, publishing, marketing, science, heritage collections and evidence.
- Generated or manipulated images may carry labelling duties under AI and media law.
Sources Filled
- Portable Network Graphics (PNG) Specification (Third Edition) - World Wide Web Consortium
- Scalable Vector Graphics (SVG) 2 - World Wide Web Consortium
- IPTC Photo Metadata Standard 2025.1 - International Press Telecommunications Council
- IIIF Image API 3.0 - International Image Interoperability Framework Consortium
- Web Content Accessibility Guidelines (WCAG) 2.2 - World Wide Web Consortium
- PROV-O: The PROV Ontology - World Wide Web Consortium
- PREMIS Data Dictionary for Preservation Metadata, Version 3.0 - Library of Congress
- Content Credentials: C2PA Technical Specification - Coalition for Content Provenance and Authenticity
- IPTC Photo Metadata Standard, International Press Telecommunications Council
- Exchangeable image file format (Exif), CIPA and JEITA
- C2PA Technical Specification, Coalition for Content Provenance and Authenticity
Open questions
- Complete coordinator-side HTTP checks and substantive source/version review without treating a successful response as claim verification.
- Develop and test pinned parent/neighbor bindings, nested schemas and mappings against the acceptance cases in boundary-design.md.
- Review specialist imaging and jurisdiction-specific use, disclosure and retention profiles before operational adoption.
- Restore independent external review before any canonical or publishable-draft promotion.
- Executable nested instance schemas, decoder implementations and standards-conformance test suites are absent.
- Pinned parent and neighbor schemas and full import/export mappings remain unresolved.
- Specialist calibration, raw formats, print profiles, multispectral data and exact Exif field mappings are deferred.
- Jurisdiction-specific rights, release, privacy, retention and publication rules require qualified adoption review.
- Animated containers are excluded except explicitly identified still extractions; multipage document semantics remain external.
- C2PA 2.2 and SVG 2 2018 are selected historical baselines; current implementation security and later revisions need adoption review.
Machine files
Provenance
world-models research · reviewable-draft
Built from: models/wm-med-005-image-graphic/spec.yaml, ver-cy/world-models/card-supplements/wm-med-005-image-graphic.json