Functional Requirements · Scene Graph · Commerce Layer · AI Pipeline
System Analysis
Comprehensive functional and non-functional requirements, system actors, Scene Graph data model, curation trust hierarchy, and integration points.
43
Functional Reqs
10
Non-Functional Reqs
6
System Actors
7
Integrations
01
Executive Summary
Scenery is a cinema discovery marketplace that transforms films into interactive, explorable cultural maps. Users search for a movie, browse an interactive scene timeline, and discover every product, song, and brand that appears within the film — all linkable to purchase or streaming.
The platform's core technical innovation is the Scene Graph: a structured knowledge graph linking Movies → Scenes → Timestamps → Elements (products, music, brands, locations). The graph is populated through a layered curation model combining AI pre-tagging, commercially-motivated shop curation, brand validation, and user contributions.
8
Core Epics
4
Curation Layers
3
Commerce Streams
1
CineCore Integration
02
System Scope & Boundaries
✓ In Scope
- Film search and discovery (movie pages, cast, director, synopsis)
- Interactive scene timeline — scrub through a film, see scene elements update
- Scene Graph: structured data model linking films, scenes, timestamps, products, songs, locations, dishes
- AI tagging pipeline: audio fingerprinting (music), computer vision (objects/products)
- Shop/retailer curation: businesses link products to specific scene timestamps
- Brand validation: official placement brands confirm and upgrade tags
- User curation: community tagging for long-tail elements
- Commerce layer: product purchase options (affiliate, official brand, local stores)
- Official movie merchandise store per film (beyond scene-specific products)
- Music discovery: song/artist info, streaming deep links, purchase options
- Watch options: rent/buy links via CineCore licensing integration
- IP licensing fee routing for merchandise sales (via CineCore)
- Filmmaker self-onboarding: filmmakers tag their own film's elements
- Location commerce: contextual booking CTAs for hotels, restaurants, and apartments linked to scenes
- Recipe pages: dishes seen in scenes link to full step-by-step recipes
- Social layer: user profiles, follow users/artists/brands, activity feed, reviews, ratings
✗ Out of Scope
- Video hosting or streaming (handled by CineCore + CDN partners)
- Rights management and revenue splitting (delegated to CineCore)
- Direct e-commerce fulfilment, shipping, or physical payment processing (bookings link out to partner systems)
- Film production tools or content creation
03
System Actors
Visitor / User
Discovers films, explores timelines, clicks products and songs, watches trailers
Registered User
Contributes tags, saves favourite films, builds watchlists, earns reputation
Shop / Retailer
Registers products, links them to scene timestamps, gains commercial visibility
Brand
Validates official placements, manages canonical product entries
Filmmaker / Studio
Registers films, self-curates scene elements, activates IP licensing
Scenery Admin
Moderates tags, manages disputes, monitors platform health
04
Functional Requirements
4.1 Film Discovery & Search
| Req. ID | Requirement |
|---|---|
| FR-01 | Users shall be able to search films by title, director, cast member, year, and genre. |
| FR-02 | Search results shall return within 200ms for queries up to 5 terms. |
| FR-03 | Each film shall have a dedicated page showing metadata, cast, director, synopsis, and watch options. |
| FR-04 | The film page shall surface an interactive scene timeline as the primary exploration element. |
| FR-05 | Related films shall be suggested based on shared cast, director, genre, and scene elements. |
4.2 Scene Graph — Data Model
| Req. ID | Requirement |
|---|---|
| FR-06 | The system shall maintain a Scene Graph structured as: Film → Scene → Timestamp → Element. |
| FR-07 | Elements shall support the following types: Product, Song, Brand, Location, Vehicle, Costume. |
| FR-08 | Each element shall carry a confidence score reflecting its curation source (AI / Shop / Brand / User). |
| FR-09 | The system shall support multiple elements of different types at the same timestamp. |
| FR-10 | The Scene Graph shall be queryable by film, scene, timestamp range, and element type. |
4.3 Interactive Scene Timeline
| Req. ID | Requirement |
|---|---|
| FR-11 | The timeline shall display scene markers across the full film duration. |
| FR-12 | As a user scrubs the timeline, the product and music panels shall update in real time. |
| FR-13 | Scenes shall display a thumbnail image, scene title, and timestamp range. |
| FR-14 | The timeline shall be usable without a film player — it operates on metadata, not video. |
| FR-15 | Timestamps shall be linkable (shareable URL with timestamp parameter). |
4.4 AI Tagging Pipeline
| Req. ID | Requirement |
|---|---|
| FR-16 | The system shall integrate an audio fingerprinting service to identify songs in film audio tracks. |
| FR-17 | The system shall integrate a computer vision model to identify objects, clothing, and vehicles in scenes. |
| FR-18 | AI tags shall be flagged with a confidence score and presented as suggestions for human validation. |
| FR-19 | The system shall provide a correction workflow where shops, brands, or admins can override AI tags. |
| FR-20 | AI processing shall be asynchronous; film pages shall be available before AI tagging is complete. |
4.5 Shop & Retailer Curation
| Req. ID | Requirement |
|---|---|
| FR-21 | Retailers shall be able to register on the platform and link their product catalogue. |
| FR-22 | Retailers shall be able to place a product at a specific film scene timestamp. |
| FR-23 | A retailer placing a product at a timestamp shall trigger the IP licensing fee calculation via CineCore. |
| FR-24 | Multiple retailers may link competing products at the same timestamp; both shall be visible. |
| FR-25 | Retailer product placement shall include a product image, title, price, and purchase URL. |
05
Non-Functional Requirements
Performance
NFR-01
Film search shall return results within 200ms at p99.
Performance
NFR-02
Scene timeline element updates shall render within 100ms of user scrub action.
Performance
NFR-03
Film page initial load shall complete within 2s on a 4G mobile connection.
Scalability
NFR-04
The Scene Graph shall support 100,000 films and 50 million element records.
Scalability
NFR-05
The platform shall support 50,000 concurrent users at peak.
Availability
NFR-06
99.9% uptime SLA for all user-facing endpoints.
Security
NFR-07
All API communication shall use TLS 1.3.
Security
NFR-08
User PII (email, location) shall be GDPR-compliant with consent management.
Accessibility
NFR-09
Platform shall meet WCAG 2.1 AA accessibility standards.
Internationalisation
NFR-10
Platform shall support English, French, Dutch, and German at launch.
06
Curation Trust Hierarchy
When multiple curation sources produce conflicting tags for the same timestamp, the following trust hierarchy is applied:
1
Filmmaker / Studio Highest
Canonical source of truth — overrides all other tags. Direct creator knowledge.
2
Official Brand Placement
Verified paid placement — overrides AI and user tags. Brand confirms their own product.
3
Shop / Retailer
Commercially motivated, high-accuracy curation. Revenue depends on correct tagging.
4
AI Auto-Tag (confidence >85%)
Reliable for well-trained categories (cars, major brands). High threshold required.
5
Community (3+ confirmations)
Elevated user tags with 3+ independent confirmations. Crowd-sourced validation.
6
Single User Submission Lowest
Enters moderation queue; displayed as unverified until approved by admin.
07
Integration Points
| Integration | Purpose |
|---|---|
| CineCore | Film licensing, watch options, IP licensing fee routing, revenue splitting |
| Spotify / Deezer / Apple Music | Music deep-link generation for streaming referrals and affiliate commissions |
| Audio Fingerprinting API | Song identification in film audio tracks (e.g. ACRCloud) |
| Computer Vision API | Object and product recognition in scene frames (e.g. AWS Rekognition) |
| Stripe | Shop subscription billing, affiliate payout processing |
| Geolocation API | Local store proximity calculation for product availability |
| Affiliate Networks | Commission tracking for outbound product and music clicks |