Beyond Unreal

Other surfaces

One character, many places to bring it to life.

An IAMX character is not tied to a single window or a single machine. The same personality, voice, memory and action set can appear as a photoreal MetaHuman inside a game, a talking head embedded in a web page, a pixel-streamed avatar that loads in any browser, or a hardened kiosk standing in a bank lobby. You author the character once — its prompt, knowledge, tools and look — and choose which surfaces to deploy it to.

This page is a tour of those surfaces: what each one is for, what it can and cannot do, and how to ship it. If you are brand new, start with the Quickstart and come back here when you know which surface you need.

Every surface below is powered by the same connection ID. Design the character once in the panel, and each surface simply attaches to it. Change the system prompt or add a tool, and all live surfaces pick it up — no re-deploy per surface.

The four surfaces at a glance

LiveFace avatar

A talking head generated from a single photo. Lightweight, embeddable, perfect for websites, support widgets and product tours. No 3D scene required.

Cloud MetaHuman

Your full photoreal character, pixel-streamed straight to the browser. Nothing to install for the visitor — they click a link and talk. Available on paid plans.

24/7 kiosks

An always-on physical station with identity verification and verified actions. Built for regulated industries — banking, telecom, public services. Enterprise.

Game NPCs

The complete Unreal experience: your character talks, looks at the player, emotes, plays body animation and calls actions inside your own project.

Same brain, everywhere

Voice, personality, knowledge bank and tool calling are shared. A surface only decides how the character is rendered and where it runs.

Mix and match

Nothing stops one character from being a game NPC, a website LiveFace and a lobby kiosk at the same time. Deploy to as many surfaces as your plan allows.

Choosing a surface

Pick by where your audience already is and how much fidelity the moment demands. Use this table as a quick decision aid.

Surface	Visitor installs	Fidelity	Best for	Plan
LiveFace avatar	Nothing — runs in a web page	Photo-realistic head & shoulders	Websites, help widgets, landing pages, social clips	All plans
Cloud MetaHuman	Nothing — pixel-streamed	Full photoreal body & scene	Premium web demos, virtual receptionists, showrooms	Paid
24/7 kiosk	N/A — dedicated hardware	Full photoreal + physical presence	Branches, lobbies, regulated self-service	Enterprise
Game NPC	Your game / app	Full photoreal in your own world	Games, training sims, interactive experiences	All plans

LiveFace avatar

LiveFace turns a single still photo into a speaking, expressive talking head. There is no rig to build, no scene to light and no engine for the visitor to run — the avatar streams into an ordinary web page. It is the fastest way to give a character a face on the open web.

What you get

Natural lip movement synced to the character's speech, driven by the same voice as every other surface.
Head motion, blinks and subtle expression so the face reads as alive rather than a frozen portrait.
A drop-in embed you can place in any site, help widget or campaign page.
Full conversational behaviour — listening, answering, knowledge lookups and tool calls all work exactly as they do in Unreal.

When to reach for it

Choose LiveFace when you want maximum reach with minimum weight: a support persona in the corner of your docs, a spokesperson on a product page, or a quick face for a character you are still prototyping. If you need a full body, a walkable scene or physical presence, step up to Cloud MetaHuman or a kiosk.

A good LiveFace photo is front-facing, evenly lit, with the whole head in frame and no heavy occlusion (hair over the eyes, sunglasses, hands on the face). The cleaner the source, the more natural the motion.

Cloud MetaHuman

Cloud MetaHuman takes your full photoreal character — the same one you build for Unreal — and pixel-streams it straight to the browser. The heavy rendering happens on our infrastructure; the visitor receives a video-quality stream and sends back their microphone and clicks. They never install anything, never download a build, and never need a capable GPU.

How it feels to the visitor

Open a link — the character loads in a normal browser tab, desktop or mobile.
Grant the mic — one permission prompt, then they can just talk.
Converse live — full-body photoreal character responds with voice, lip sync, gaze and gesture, streamed in real time.

Why pixel streaming

A photoreal character is expensive to render. Pixel streaming moves that cost off the visitor's device, which means:

Zero install — no launcher, no plugin, no multi-gigabyte download.
Consistent quality — every visitor sees the same fidelity regardless of their hardware.
Instant updates — change the character in the panel and the next session reflects it.

Cloud MetaHuman is available on paid plans, because each live session holds real rendering capacity for the duration of the conversation. Concurrency (how many visitors can talk at once) scales with your plan — see plans & pricing.

Prototype on LiveFace, then promote to Cloud MetaHuman when the character is ready for a premium moment. Because both surfaces attach to the same connection ID, there is nothing to re-author — you are switching the renderer, not rebuilding the character.

24/7 kiosks

The kiosk surface is a character that stands in the physical world and never logs off. It is built for regulated, high-trust self-service: a customer walks up, the character greets them, verifies who they are, and can then perform sensitive actions on their behalf — all with a photoreal face and a natural voice instead of a touchscreen menu tree.

Identity verification

Before a kiosk performs anything sensitive, it can confirm the person in front of it is who they claim to be. The flow combines a document read with a live face check and only proceeds when both agree.

Walk-up
   │
   ▼
[ Greet ]  ── face seen, not yet identified
   │
   ▼
[ Verify identity ]  ── document + live face compared
   │
   ├── match ──────────►  [ Verified session opens ]
   │                          verified actions unlocked
   ├── mismatch ───────►  [ Politely declined ]
   └── no live face ───►  [ Ask customer to step into view ]

The kiosk uses operator-customizable prompts for every stage — the greeting for a known vs. unknown face, the message on a successful match, on a mismatch, on a missing live face, and the welcome-back message when a already-verified customer returns within the session window. You control all of that copy without touching code.

Verified actions

Once a session is verified, the character can call gated tools — actions that are only allowed to fire when identity has been confirmed. This is what separates a kiosk from a talking FAQ: it can actually do the thing the customer asked for, safely.

Sensitive tools stay locked until the verification gate passes.
Each action carries the verified customer context so your backend can trust the request.
If a customer asks for a gated action without being verified, the character declines gracefully using your configured message.

Kiosk identity verification and verified actions are an enterprise capability, intended for deployments with a real-world duty of care (banking, telecom, public services). Talk to us through the panel to enable it for your organisation.

Designed for always-on

Kiosks assume nobody is watching the machine. They greet on approach, recover cleanly between customers, ignore their own audio so they don't answer themselves, and hold a short verified-session window so a returning customer isn't asked to prove themselves twice in a minute.

Game NPCs

The game NPC surface is the complete experience running inside your own Unreal project. This is the surface the plugin was born for: a photoreal character living in your world, aware of the player, and reacting in real time.

Talking

Streaming conversation with natural voice and precise lip sync, driven by GPT, Claude, Gemini or a local Ollama model.

Looking

Eye and head gaze that tracks the player during conversation, so the character feels present instead of staring through you.

Emoting

Facial expression and body animation — idle, listening and talking states blend automatically as the conversation flows.

Acting

Tool and action calls let the character trigger events in your game — open a door, hand over a quest, hit an external API.

Perceiving

Optional vision lets the character see from its own viewpoint or the player's, and reason about what's in front of it.

Remembering

Knowledge bank lookups and long-term memory give the NPC continuity across a conversation and across visits.

How it fits your project

Drop the component onto a character, point it at a connection ID, and wire the delegates you care about (for example OnActionTriggered) in Blueprint. Everything else — voice, lip sync, gaze, body animation — is handled for you. The full node and delegate reference lives in the Blueprint API docs, and setup is covered in Install.

Unlike Cloud MetaHuman, a game NPC renders on the player's own machine inside your build. That means no streaming session cost — the trade is that the player runs your project. Use game NPCs for shipped experiences, and Cloud MetaHuman when you need zero-install reach on the web.

One character, every surface

The single most important idea on this page: a surface is a destination, not a new character. The personality, the voice, the knowledge, the tools and the memory all belong to the connection ID. Surfaces just decide how that character is rendered and where it runs.

                 ┌─────────────────────────────┐
                 │   Character (connection ID) │
                 │  prompt · voice · knowledge │
                 │      tools · memory         │
                 └──────────────┬──────────────┘
                                │  (author once)
        ┌───────────────┬───────┴───────┬────────────────┐
        ▼               ▼               ▼                ▼
   LiveFace        Cloud            24/7             Game
   avatar          MetaHuman        kiosk            NPC
   (web embed)     (pixel stream)   (hardware)       (your Unreal build)

Edit the system prompt, add a knowledge file, or define a new tool once — and every live surface reflects it. That's what makes IAMX design once, deploy everywhere: you invest in the character, not in four separate integrations.

Next steps: build your character in the panel, try it instantly as a LiveFace or Cloud MetaHuman on the live demo, then follow Quickstart to drop it into Unreal as a game NPC. Questions and showcases live on the forum, and release notes are on GitHub.