> ## Documentation Index
> Fetch the complete documentation index at: https://docs.zevor.ai/llms.txt
> Use this file to discover all available pages before exploring further.

# Build an AI influencer

> From zero to your own AI influencer: consistent face, different looks, vertical videos with the same identity. Full recipe with credits and timing.

An AI influencer is an AI character that keeps **the same face** across
many videos and posts. The recipe: generate the face once, save it as
an avatar, reuse it on every generation.

## Outcome

* 1 saved avatar (fixed face).
* 3-5 photos of the avatar in different scenes (outfits, locations).
* 1-2 vertical 9:16 videos ready for TikTok or Reels.

## Time and credits

* **Total time**: \~30-45 minutes.
* **Credits**: \~80-120 cr.

Fits in a **Starter** month (120 cr), or a \$4.99 top-up pack (30 cr)
if you already have the avatar created.

## Steps

<Steps>
  <Step title="Generate the base face">
    In **Create → Characters**, describe the character with the chips
    (ethnicity, age, hair, distinctive marks) or write your own free
    prompt. Generate 4 candidates and keep the one you like best.

    Cost: 4 cr × 4 candidates ≈ 16 cr.
  </Step>

  <Step title="Save as avatar">
    On the result card, **Save** + name. It lands in your avatar
    library and is reusable in any mode.
  </Step>

  <Step title="Variations (Remix)">
    From the avatar's card, **Remix**: change outfit, scene, pose.
    Three variations are plenty for a first series. The face stays.

    Cost: \~12 cr per 4-image variation.
  </Step>

  <Step title="Take the avatar to video">
    Two paths:

    * **Image to video**: use one of the variations as the base image
      and describe the motion (8 s, Pro with audio = 64 cr).
    * **Motion**: film a clip of yourself moving, upload the avatar
      as the character image, and the avatar copies the movement
      (Motion standard 6 cr/s).
  </Step>

  <Step title="Talking photo (optional)">
    If you want the avatar to speak, take it to
    [Talking photo](/en/video-tools/talking-photo) with your audio or
    TTS. 4 cr per second of speech.
  </Step>
</Steps>

## Best practices

* **One face, many contexts**. The strength of an AI influencer is
  that the audience recognizes the face across posts. Change the
  scene, not the identity.
* **Iterate on Lite**. Find the prompt on Lite, then bump up to Pro
  or 4K for the hero piece.
* **Front or 3/4 shot** for video and for Talking photo: the output
  is much cleaner.
* **Save the prompt** that worked (it's in your history). Saves time
  on future series.

## Common mistakes

* **Changing the face every time**: if you describe the character
  from scratch on every generation, it comes out different. Create
  **one** avatar and reuse it.
* **Overlong prompts**: a dense description can contradict itself.
  Short sentences, one detail per sentence.

<Note>
  Visual walkthrough at [zevor.ai/characters](https://zevor.ai/characters).
</Note>