How It Works

Scan once.
Consent once.
Authorize everywhere.

DIAP is the consent router between talent and every studio, platform, and AI pipeline in the world. Here's how the entire system works — from capture to authorization to the open ecosystem.

The Core Model

DIAP is infrastructure, not a capture device

Studios bring their own scanners. Talent brings their identity. DIAP sits in the middle and makes consent enforceable.

DIAP doesn't capture faces, record voices, or train models. That's the whole point — the protocol exists because other people are doing that without consent.

Instead, DIAP stores consent rules and mathematical references (one-way embeddings that can verify identity but can't reconstruct it). The database holds math, not faces. If someone breaches the vault, they get hashes that are useless without the original source material.

Studios use their own capture equipment. Talent registers once. Every studio in the world checks with DIAP before using anything. The talent never sits for the same scan twice.

End-to-End Flow

1.Capture happens anywhere

Studio, agency, or self-service — any DIAP-Certified tool captures identity data (face, voice, motion). Raw data stays with the capture source.

2.Embeddings extracted

The certified tool converts raw data into standardized one-way embeddings — mathematical fingerprints that verify identity but can't reconstruct the original.

3.Anchored in the vault

Embeddings are submitted to the talent's Identity Vault on DIAP. Raw captures are discarded after extraction. The vault holds math, not biometrics.

4.Consent rules set

The talent (or their agent) sets two-layer consent: Layer 1 controls who can even see they exist in the registry. Layer 2 controls per-project usage rights.

5.Studios request access

When a studio wants to use a talent's identity, they request a License Token from DIAP. DIAP checks consent and either grants or denies — scoped, time-limited, auditable.

6.Render receipts close the loop

After rendering, the studio submits a Render Receipt — cryptographic proof of what was produced, under what authorization. The audit trail is complete.

Capture to Vault

Three paths into the Identity Vault

Whether the talent is already registered, has an agent, or has never heard of DIAP — the system handles every scenario.

The simplest case. The actor already registered on the DIAP platform. They have a DIAP ID. The studio knows this because they checked before production, or the talent's agent provided it on the call sheet.

1
Studio's DIAP-Certified tool generates Identity Anchor Payload
2
Studio calls POST /api/v1/identity/anchor/submit with the talent's DIAP ID
3
DIAP holds the anchor in PENDING state — never auto-approved
4
Talent receives notification: "Studio XYZ submitted a FACE_3D anchor. Approve?"
5
Talent approves → anchor added to vault. Rejects → payload deleted, studio notified.

The talent is always in control. The studio cannot dump data into someone's vault without explicit approval.

Registration

Three ways to register identity

Quality varies. Consent model is identical. Every tier produces standardized DIAP embeddings.

Self-Service

Phone camera + mic

QualityBasic
Cost to talentFree
Use caseIndie creators

Enough to establish identity anchors and start managing consent. Upgraded when working with certified tools later.

Agency-Assisted

DIAP-Certified capture station

QualityProfessional
Cost to talentCovered by agency
Use caseRepresented talent

Agencies set up certified stations for their roster — a competitive advantage. "All our talent are DIAP-registered."

Studio-Captured

LiDAR, photogrammetry, voice booth

QualityProduction-grade
Cost to talentPart of production budget
Use caseFilm & VFX production

Captures happen during production anyway. DIAP standardizes the output and routes it to the talent's vault.

Delegation

The manager and agent model

Built for how the entertainment industry actually works. Agents manage. Talent retains ultimate control.

Permission
Talent
Agent
Studio
Submit identity anchor
Approve anchor into vault
If delegated
Set Layer 1 visibility
If delegated
Approve Layer 2 license
If delegated
Revoke consent (kill switch)
View audit log
Own talent
Own submissions
Delete identity data

The kill switch never gets delegated.

An agent can manage everything day-to-day — approvals, visibility, licensing — but only the talent themselves can revoke all consent and delete their identity data. This is a non-negotiable trust principle. No contract, no power of attorney, no corporate structure can override it.

Economics

Who pays for what

DIAP doesn't charge for scanning — it charges for authorization. The capture tools are a commodity layer. The consent layer is DIAP's value.

Studios already pay for face scans, voice recordings, and motion capture as part of production. DIAP doesn't add a new cost — it standardizes the output format and routes the embeddings to the talent's vault instead of the studio's proprietary database.

No studio can feel like a free-rider because nobody "uses the scan." Everyone uses DIAP's authorization layer, and everyone pays for that access equally. The scan is the talent's property. The authorization is DIAP's service.

Identity capture (face, voice, motion)
This is the talent's asset — like professional headshots
Talent, agency, or production budget
Vault registration & consent management
Maximum adoption. Charging talent kills the network.
Free for talent — always
License Token requests
This is where DIAP monetizes. Every studio pays for access.
Studio — per API call
Render Receipts
Audit trail cost sits with the entity doing the rendering.
Studio — per receipt
DIAP Certification
Paying for trust, compliance verification, and network access.
Studio or tool maker — annual fee
Ecosystem

DIAP provides the reference app. Anyone can build on top.

If DIAP is the only app, the ecosystem can't grow faster than one engineering team. The protocol must be open.

Think of it like email. Gmail is Google's email app, but Outlook, Apple Mail, and Thunderbird all access email through the same protocols. Google doesn't block them. The protocol is what matters, not the app.

DIAP provides the reference web platform and the API. Anyone can build a client on top — talent agencies, unions, studios, independent developers. The data stays in DIAP's vault. The consent logic follows DIAP's rules. But the interface can be anything.

Talent Agencies

"CAA Identity Manager"

A branded app where agents manage their entire roster's consent through their own interface. All data lives in DIAP's vault.

Unions & Guilds

SAG-AFTRA Member Portal

Members manage DIAP consent alongside union contracts. The union app calls DIAP's API — they don't rebuild the consent engine.

Studios & Production

ShotGrid / ftrack integration

DIAP checks embedded into production software. When a VFX supervisor starts a shot, the system auto-verifies the License Token.

Independent Developers

Mobile-first talent app

Better UX, social features, portfolio display — specialized for indie creators. Connected to DIAP's vault through the API.

What a third-party app needs

From DIAP (we provide)

  • Developer account & API credentials
  • SDK & API documentation
  • OAuth/OpenID Connect integration
  • DIAP certification (for sensitive operations)

What they can't do

  • Store embeddings or identity data locally
  • Bypass the consent model
  • Issue their own tokens or receipts
  • Access the kill switch on behalf of talent
Security

How the open ecosystem stays secure

Security comes from controlling the protocol layer, not the app layer.

OAuth/OpenID Connect authentication

Talent logs in through DIAP's identity provider. Third-party apps get a scoped access token, never the user's credentials. Talent can revoke any app's access instantly.

Scoped API permissions

Developers declare what permissions they need (read profile, manage consent, submit anchors). Talent grants scopes per app. A status-check app doesn't get consent management access.

Certification tiers

Read-only apps need minimal registration. Talent management apps require full security audits. Capture tools need additional embedding-quality audits. The trust requirements scale with the risk.

Complete audit trail

Every API call from every third-party app is logged. Talent sees which app did what, when. If an app misbehaves, DIAP revokes its credentials instantly across all users.

No data residency in third-party apps

The golden rule. Apps can display identity data but must never cache or store it beyond the current session. Identity data lives in DIAP's vault — period.

Integration Levels

Three types of tools that build on DIAP

From face scanners to AI pipelines to entirely new identity modules — the protocol is extensible.

Type 1 — Capture Tools

Face scanners, voice recorders, motion capture rigs — the front-end devices that create initial biometric data. They're not part of DIAP itself; they're certified integrations.

Must follow DIAP embedding formatSubmit to vault API, not local storageDelete raw captures after extractionDIAP-Certified required

Type 2 — Rendering & Generation Tools

The AI pipelines that actually use identity data — deepfake generators, voice synthesizers, motion retargeting systems, script derivative engines.

Request License Token before processingVerify authorization via referenceSubmit Render Receipt when doneNever touch vault directly

Type 3 — Custom Modules

DIAP defines identity modules for voice, face, expression, motion, and script. But the protocol is extensible — hand geometry, gait patterns, dental records for forensic VFX, or any new asset type can be proposed and certified.

Must use DIAP consent modelRegister through Trust RegistryFollow token & receipt formatPass certification for interoperability

Like USB — anyone can make a device, but it has to follow the spec to work with the ecosystem.

Developer Economics

Costs for building on DIAP

Talent is always free. Developers pay proportional to the value they extract.

Register as developer, get API keysFree
Read-only API (verification, public registry)Free (rate-limited)
Talent management API (consent, anchors, audit)Usage-based per API call
DIAP Certification (sensitive operations)Annual fee
Trust Registry listingIncluded with certification

A small indie developer building a portfolio app for talent pays almost nothing — low API volume. Adobe integrating DIAP into their enterprise pipeline pays significantly more — high API volume plus enterprise certification.

The pricing scales with value extracted, not with the number of apps. This is the same model that turned Stripe into a $95B company — they didn't build every checkout page themselves. They built the payment infrastructure and let a million developers build on top.

Data Architecture

What lives inside the Identity Vault

The vault stores math, not faces. Multiple anchors per module, quality upgrades over time.

// Talent's Identity Vault
FACEmodule
Anchor #1 BASIC
Self-service · Phone camera · 2026-01-15 · ACTIVE
Anchor #2 PRODUCTION
Studio LiDAR · Studio XYZ (cert_abc) · 2026-03-19 · PENDING APPROVAL
VOICEmodule
Anchor #1 BASIC
Self-service · Phone mic · 2026-01-15 · ACTIVE
MOTIONmodule
No anchors registered yet
Verification requests match against highest-quality available anchor. Basic anchors are supplemented, never replaced.
Usage Tracking

Every use is logged — even within an existing authorization

Studios don't re-request for every session within an approved scope. But every single use — especially AI modifications — is tracked and visible to the talent.

When a studio gets a License Token for a talent's voice on a project, they can work freely within that scope — recording sessions, AI-assisted line adjustments, quality fixes — without interrupting production with repeated requests. But DIAP logs every interaction separately.

One token. Multiple Render Receipts. Full transparency on every use.

// Real scenario: Voice authorized March 8, AI-modified March 16
Mar 8
LICENSE ISSUED
Studio XYZ — VOICE for "Project Zenith Ep.3"
Token: lt_voice_001 · Rights: VOICE_RENDER, VOICE_DUBBING · Expires: Jun 8
Mar 8
RENDER RECEIPT #1
247 voice lines recorded and processed
Output: VOICE_RENDER · Duration: 42min
— 8 days pass —
Mar 16
TOKEN VALIDATED ✓
Same token lt_voice_001 — still valid (84 days remaining)
No new request needed. Same project, same scope.
Mar 16
RENDER RECEIPT #2 — AI MODIFIED
14 lines AI-modified for "Project Zenith Ep.3"
Model: studio_voice_synth_v2 · Source: Receipt #1 · Modification: 5.7% of content

AI Modification Flags

When AI is involved in modifying previously-authorized content, the Render Receipt captures exactly what happened — so the talent sees not just that their voice was used again, but how.

modification_type
AI_MODIFIED
Flags that AI was involved
original_receipt_ref
rr_001
Links back to original recording
ai_model_used
studio_voice_synth_v2
Which AI model performed it
modification_ratio
0.057 (5.7%)
Exactly how much was changed

Requires new License Token

  • Different project or episode
  • Different rights (render → training)
  • Token expired
  • Talent revoked consent
  • Different identity module

No new request needed (still logged)

  • Same project, same scope, same rights
  • AI modification of authorized content
  • Multiple sessions within token validity
  • Quality adjustments on same deliverable
Enforcement

Three mechanisms that make compliance the only rational choice

No system can put a camera inside every render farm. But DIAP makes non-compliance extremely risky and highly detectable.

Mechanism 1 — Render Receipts

Mandatory, after every render

The primary enforcement tool. Studios must submit a Render Receipt when producing output using a talent's identity. This is a contractual and technical requirement of being DIAP-Certified. DIAP validates the token is still valid, scope matches, time window is respected, and logs everything to the talent's audit trail.

// What a Render Receipt contains
license_token: "lt_abc123"
output_type: "VIDEO_FRAME_SEQUENCE"
output_hash: "sha256:..." ← integrity proof
frame_count: 2847
watermark_id: "wm_def789"
model_used: "internal_face_gen_v3"

Mechanism 2 — Token Heartbeat & Scoped Expiry

Real-time monitoring during active sessions

License Tokens are short-lived and scope-locked. Studios must refresh them periodically. Active rendering sessions send heartbeat pings. If heartbeats stop but no Render Receipt is submitted — that's a compliance flag.

ISSUED
ACTIVE
HEARTBEAT
HEARTBEAT
RECEIPT
CLOSED
Every step logged.·Missing heartbeat + no receipt = compliance flag ⚠️

Mechanism 3 — Watermark Verification

Passive, ongoing, anyone can check

Every output rendered under DIAP authorization includes dual-layer watermarks: a pixel-level steganographic payload (invisible to the eye) and an ultrasonic audio frequency fingerprint (inaudible at 18–20.5kHz). Each asset gets a unique Sound ID. When content appears anywhere — streaming platforms, social media, advertisements — anyone can verify it through either channel.

Valid watermark

Linked to active token and matching receipt → Authorized ✓

Expired/revoked watermark

Token no longer valid → Flagged for review ⚠️

No watermark

Content using identity without DIAP authorization → Unauthorized ✗

The honest gap — and why compliance is still rational

No system can track what happens on hardware you don't control. If a studio renders frames internally and never submits a Render Receipt, DIAP doesn't see it in real-time. But the enforcement mechanisms make non-compliance extremely risky:

Certification audits

Periodic audits compare studio rendering logs with DIAP receipt records. Gaps are violations.

Dual-layer watermark detection

Unauthorized content without valid watermarks (pixel or audio frequency) is detectable by anyone running verification.

Token expiry

No refresh call = no active authorization. Any output is unauthorized by definition.

Industry pressure

Distributors and streamers can require DIAP verification before accepting content.

Same model as financial auditing — you make the consequences severe and detection likely enough that compliance is the rational choice.

The protocol is the product.

DIAP doesn't capture your face. It doesn't train on your voice. It doesn't build the tools. It makes sure everyone who does has to ask first — and prove they did.

DIAPProtocol v1.3

Digital Identity Authorization Protocol · How It Works