HUANTU

First call / about 5 minutes

Your first protected call, end to end.

Install Huantu, create a persona from one clear portrait, activate it, and select Huantu Camera in the call app you already use.

Download for Windows Start the guide
Steps05
Persona setupName + face
Call deviceHuantu Camera
Media uploaded0 B

01 / QUICK START

From install to caller output.

Follow these steps in order the first time. Later calls usually begin at step three.

  1. 01

    INSTALL + SIGN IN

    Let Huantu prepare the local engine.

    Install Huantu and sign in with Google or email. The first run downloads about 970 MB, then expands the local engine to about 1.5 GB. Interrupted downloads resume automatically.

    Current testing release Windows may show an "Unknown publisher" warning while production signing is pending. You do not need to install Python, Node.js, CUDA, or OBS.
    FIRST RUN
    • 01 Hardware checked READY
    • 02 Local engine prepared READY
    • 03 Huantu Camera installed READY
  2. 02

    CREATE A PERSONA

    Add a name and one face portrait.

    Open Personas and choose Create persona. Enter a short name, continue to Face, then choose a JPG, PNG, or WebP image containing one person.

    1. 1 Enter a name you will recognize before a call.
    2. 2 Choose the portrait and wait for "One face confirmed."
    3. 3 Confirm you have permission to use the image.
    4. 4 Select Create persona.
    CREATE A PERSONA
    12
    ONE FACE CONFIRMED
  3. 03

    SELECT THE PERSONA

    Activate the identity you want to use.

    Find the persona card in your identity library and press Use face. An active banner confirms which saved identity Huantu will use. You can switch to another saved persona before a call.

    Look for The card button changes from "Use face" to "Active."
    JLOCAL PERSONA

    JordanFace checked

  4. 04

    START THE SESSION

    Check the source, webcam, and caller view.

    Open Session. Confirm your persona under Source identity, leave Webcam on auto-detect or select the correct camera, and begin with the Balanced quality preset.

    Press Start call output and wait for the preview to show Caller output live. Caller view is the unmirrored image sent to the other app; Mirror me changes only your local preview.

    SESSION READINESS
    Identity
    Jordan
    Camera
    Auto-detect
    Quality
    Balanced
    Output
    LIVE
  5. 05

    CONNECT THE CALL APP

    Select Huantu Camera.

    Open the camera settings in Zoom, Google Meet, Discord, Telegram, OBS, or another calling app and select Huantu Camera. Keep Huantu open while the call is using its virtual camera.

    When the call ends, return to Huantu and select Stop call output.

    If Huantu Camera is missing Use Install virtual camera or Repair in the Huantu footer, approve the Windows administrator prompt, and reopen the call app.
    REAL CAMERA
    LOCAL
    HUANTU
    SELECT
    CALL APP

02 / BETTER RESULTS

Choose a strong portrait.

Source and live-camera angles matter. A clean, frontal image gives Huantu less to guess.

USE

Clear, frontal, evenly lit

  • One person facing mostly forward
  • Neutral expression and even light
  • Forehead, eyes, cheeks, and chin visible
  • A sharp image without heavy filters

AVOID

Crowded, cropped, or obscured

  • Group photos or another face in frame
  • Side profiles, sunglasses, or coverings
  • Tight crops removing chin or forehead
  • Strong shadows, blur, or a tiny face
01Camera at eye level
02Sit roughly 60–90 cm away
03Face fills about 20–30% of frame

03 / PC REQUIREMENTS

Know the performance floor.

Huantu checks the PC automatically before opening the workspace.

Component Minimum Recommended
Windows Windows 10 version 2004 or newer, 64-bit Windows 11, 64-bit
Processor 64-bit CPU with 4 logical cores 8 or more logical cores
Memory 16 GB RAM 32 GB RAM
Graphics DirectML-capable discrete GPU with 8 GB dedicated VRAM, RTX 3060-class or better RTX 4060-class or better with 8 GB+ dedicated VRAM
Storage 8 GB free space 15 GB free SSD space

04 / SAFE BY DEFAULT

The persona never jumps to someone else.

The first person in view owns the active face lock. Additional people are ignored. If Huantu loses the locked person, caller output goes black instead of moving the identity to another face.

LOCKED PERSONTRANSFORMED
ADDITIONAL FACEIGNORED
LOCK LOSTBLACK OUTPUT

05 / QUICK HELP

If something is not ready.

Most first-call issues have a visible action inside Huantu.

START BUTTON DISABLED

Select an identity first.

Choose a saved persona under Source identity and wait for the face engine to finish preparing.

HUANTU CAMERA MISSING

Install or repair it.

Use the footer action, approve the administrator prompt, then reopen the camera list in your call app.

OUTPUT IS BLACK

Return the locked person to view.

Use one well-lit face, stay in frame, and check Caller view. Black is the protected fallback when tracking is lost.

LOW FRAME RATE

Reduce GPU pressure.

Close GPU-heavy apps or games, choose the Performance preset, and keep the webcam at a normal call resolution.

VOICE

Real-time voice identity is marked Coming soon. Face personas and protected video output are available now.

READY / WINDOWS

Create your first persona.

Start with a clear portrait you have permission to use.

Download Huantu Back to Huantu home