Skip to content
Truly Imagined Truly Imagined Docs

Building your model

If you’re an actor generating as yourself, you first build your model — the reference data your likeness is generated from. It’s a guided flow in the Studio: enable usage, then complete as much or as little of the capture kit as you like. You can record each take in the browser, upload an existing file, or choose one from your Media Vault.

Open it from the Generation Ready marker in the Studio header, or from Account → Model.

Enable usage

Section titled “Enable usage”

Before any capture, you confirm two consents:

  • Voice replication — permission to generate speech in your voice.
  • Synthetic media — permission to generate talking-head video of your likeness.

This turns on self-generation. You can withdraw consent at any time — see Consent & compliance.

Three models

Section titled “Three models”

Your account shows three models, each for a different kind of output. Setting up any one makes you generation-ready; add the others to widen what you can create.

Voice Model — audio only

Section titled “Voice Model — audio only”

For voiceover, audiobooks and podcasts. Complete the performed audio take (read the “Arthur the Rat” passage) — a short, clean clip becomes your canonical voice reference. A casual audio take is optional and improves range.

Video Model — film & cinema

Section titled “Video Model — film & cinema”

For talking-head video. Complete the reference photo (your face likeness); performed and casual video takes are optional enrichment. Talking-head generation pairs your Video model with your Voice model.

3D Model — gaming & animation

Section titled “3D Model — gaming & animation”

A 3D body replica built with Gaussian Splatting. Choose Instant (one full-body photo, a fast draft) or Premium sweep (a slow 360° video around your body, higher fidelity — on Pro/Studio plans). A 3D scan requires biometric consent; the raw capture is deleted once reconstruction completes, and you can preview and delete your avatar at any time.

Expressions

Section titled “Expressions”

An optional Expressions set teaches your model your range for richer Video and 3D likeness. You’re prompted through eight guided faces — neutral, happy, laughing, sad, angry, surprised, scared and disgusted — each with an on-screen emoji guide and a face-framing outline.

The capture experience

Section titled “The capture experience”

Recording in the browser is guided to get clean, usable takes:

  • Self-paced teleprompter for the performed reads — it scrolls at a natural reading pace, and you can pause/resume, go slower/faster, change text size, mirror the text, or simply scroll it yourself to match your pace.
  • 3-2-1 countdown before a photo snaps or recording starts, so you have time to get set.
  • Face-centering oval and a lighting check that warns if you’re too dark or too bright.
  • A live microphone level meter for voice takes, with “too quiet” and “clipping” hints.

Upload or choose from your Vault

Section titled “Upload or choose from your Vault”

Prefer to use material you already have? On any take you can Upload a file from your device or pick one from your Vault (a headshot, or previously recorded audio/video for your reel). Because uploaded material skips live capture, you confirm a short rights attestation — “I own or have the rights to use this material” — before it’s accepted. Live recordings don’t need this.

Everything you capture is filed to your My AI Model Data folder in the Media Vault.

Readiness at a glance

Section titled “Readiness at a glance”
  • The Generation Ready marker in the header is green once at least one model is set up, and a prompt to finish setup otherwise. Click it to jump into the flow.
  • Account → Model opens a condensed status — Voice · Video · 3D, each ON or OFF — with a Set up shortcut for anything missing and Manage to add more data.
Section titled “Related”