Technology · Embedded capability

Facial liveness, matched to the risk.

Use 3D when the highest real-time proof of presence matters. Use 2D when you want strong liveness with a much faster, easier experience. Axon puts the right mode inside the identity journey - not beside it.

Design a Liveness Policy Compare 2D and 3D

Virtually unbreakable

FaceTec 3D · $600,000 spoof bounty

Very resistant

2D liveness · fast, passive, offline-ready

The short answer

The right liveness method is the one that matches the threat, channel and consequence of getting it wrong.

Facial liveness is not a product we usually put on a shelf by itself. It becomes valuable when it is combined with document trust, face matching, device intelligence and a business decision inside an identity verification flow.

That is why Axon supports both 2D and 3D. 2D gives a strong, fast and easy liveness signal for journeys where usability matters. 3D adds the strongest live-session assurance for journeys where the extra time, online dependency and capture controls are justified.

Identity verificationSIM registrationAccount recoveryAssisted enrolmentRetrospective review

Two modes, one policy

3D and 2D both have a place.

2D is not a weak option; it is the faster, easier option. 3D is the highest-security option. The right answer depends on the risk and the journey.

Spoof-bounty backed

3D liveness

A guided online video capture builds depth and motion evidence across the live session. FaceTec positions its 3D Liveness as virtually spoof-proof and backs that security claim with an active $600,000 Spoof Bounty Program.

$600,000

Spoof bounty

“Virtually spoof-proof” 3D security

FaceTec’s public programme invites real-world attacks across presentation and injection vectors.

Input

Live guided video capture

Assurance

Virtually unbreakable — $600K bounty

Experience

Guided selfie, more controlled

Processing

Online real-time flow

Best fit

Remote onboarding and account recovery
SIM swap and other high-risk transactions
Watchlist enrolment and trusted re-authentication
Journeys exposed to replay, deepfake or injection attacks

Trade-off: maximum security comes with more UX friction. 3D is slower, online-first and more sensitive to lighting, camera quality and capture conditions.

Maximum reach

2D liveness

Passive analysis assesses a single frame for presentation-attack cues. It is a strong liveness signal in its own right: much faster, easier for the user and near 3D assurance for many lower-friction journeys.

Input

Single image or frame

Assurance

Very resistant to spoofing

Experience

Near-instant, passive

Processing

Edge, offline or server

Best fit

Older images and previously captured enrolments
Offline or intermittently connected operations
High-volume, low-friction customer journeys
Fast customer verification and batch risk screening

Trade-off: 2D is not quite as strong as a live 3D session, but it is faster, simpler and usually less sensitive to lighting and camera conditions.

A note on historical images. 2D analysis can flag visual spoof indicators in an older image. It cannot recreate missing session evidence or prove that the subject was live at the original moment of capture.

Decision guide

Choose by journey, not by slogan.

The same organisation can use both modes. Policy selects the right one from transaction risk, usability needs, connectivity and the age of the input.

Adaptive by design

Start with strong, low-friction 2D, then step up to 3D when the risk score, transaction type or previous evidence justifies the added friction.

High-risk digital onboarding

Bind the identity decision to a fresh, controlled live capture.

Fast, easy customer journey

Keep the experience near-instant while retaining a strong anti-spoof signal.

Offline field enrolment

Score locally where supported, then synchronise the decision evidence.

2D / adaptive

Existing image archive

Assess stored images for visual spoof cues without recapturing the subject.

Very high-volume, low-friction flow

Prioritise throughput, fewer retries and an almost invisible customer experience.

Privileged recovery or SIM swap

Use the stronger live-presence signal where account takeover risk is high.

Threat model

Liveness is more than spotting a printed photo.

Modern identity attacks target the image, the live session and the capture path itself. A credible design considers all three.

Attack vector

What the engine looks for

Printed photo or cut-out

2DStrong signal

3DStrongest

Texture, boundary, reflection and depth cues.

Screen or video replay

2DUseful signal

3DStrongest

Display artefacts plus session and motion evidence.

2D / 3D mask

2DVariable

3DStronger

Effectiveness depends on material, quality and capture conditions.

Synthetic face / deepfake

2DRisk signal

3DStronger

Live multi-frame evidence raises the cost of a convincing attack.

Camera or video injection

2DLimited alone

3DCan address

Requires protection around the capture path, not only image analysis.

Standards are a baseline

PAD testing is useful evidence, but no single certificate covers every evolving attack.

Protect the capture path

Device integrity, replay controls and injection resistance matter alongside the model.

Decide what happens next

A score needs policy: pass, retry, step up, refer for review or stop the journey.

Built in, not bolted on

One decision inside the identity flow.

Axon integrates capture, liveness, matching and business rules so the customer sees one coherent journey and the organisation receives one auditable outcome.

Capture

App, web, agent or COMET device

Policy

Risk selects 2D, 3D or step-up

Decision

Liveness + match + device signals

Workflow

Pass, retry, refer, stop or review

Identity Verification

Liveness, document capture, NFC and face matching combined into one decision-ready onboarding flow.

Polaris

Risk-based 2D or 3D liveness inside SIM registration, subscriber updates and higher-risk SIM lifecycle events.

Assisted & COMET Workflows

A mode selected for the device, operating environment and connectivity available at the point of service.

Start with the journey

Let us map the right liveness policy to your channels.

We will work from fraud risk, connectivity, throughput, lighting and camera conditions, and customer experience - then embed the right mode into the solution.

Talk to the Identity Team

Common questions

Facial liveness, without the fog.

Clear answers for product, fraud, security and operations teams.

What is facial liveness detection?+

Facial liveness detection estimates whether a face presented to an identity workflow belongs to a real person participating in the capture, rather than a photograph, screen replay, mask, injected video or synthetic face.

Is 3D liveness always the right choice?+

No. 3D provides the strongest live-session evidence and is preferred for very high-risk remote journeys. 2D is also very strong, and is often the better product choice when speed, ease, offline operation, a single stored image or minimal user friction matters more.

Can 2D liveness be run on an existing image?+

Yes. A 2D engine can assess a previously captured image for visual presentation-attack cues. It cannot retroactively prove that the person was live when the image was originally captured, so the result should be treated as a risk signal rather than fresh proof of presence.

Can liveness run offline?+

2D liveness can be deployed on-device or at the edge for offline and intermittently connected workflows, subject to the selected engine and device profile. 3D is normally an online, integrated real-time capture flow, so it carries more UX friction and is more sensitive to lighting, camera quality and capture conditions.

Are passive, active, 2D and 3D the same thing?+

No. 2D and 3D describe the facial evidence being analysed. Passive and active describe the user experience. A passive flow asks for no deliberate action; an active flow asks the user to complete a challenge. Axon selects both the evidence type and interaction model for the journey.

Does Axon sell liveness as a standalone product?+

No. Liveness is delivered through IDToolbox, part of the Identity Verification solution, rather than sold on its own. That lets the liveness result work alongside document trust, face matching, device intelligence and business rules instead of becoming an isolated score.

Is ISO 30107-3 certification enough?+

It is a useful presentation-attack testing baseline, not a complete security strategy. A production design also needs to consider camera and video injection, synthetic media, device integrity, replay controls, model updates and the business response to an uncertain result.