> ## 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.

# Recreate a viral clip with Motion

> Film a clip of yourself moving, upload an avatar as the reference, and Motion gives you back the avatar doing the same movement. Step-by-step recipe.

Motion copies the movement of your clip (or any reference) onto the
face/body of an avatar or a photo. It's the recipe behind those "this
AI character dancing like X" videos.

## Outcome

* One vertical 9:16 video with your avatar (or any photo of a person)
  doing the same movement as your reference clip, keeping the
  original audio.

## Time and credits

* **Total time**: \~5-10 minutes.
* **Credits**: 18-48 cr for a 3-8 s clip on Motion standard.

## Steps

<Steps>
  <Step title="Film the motion clip">
    3-30 s of yourself moving or talking, with good light and your
    face visible if you want it in the result. Steady camera, close
    or medium shot. Any phone works.
  </Step>

  <Step title="Prepare the character image">
    A vertical image of the character that will "perform" the clip.
    It can be:

    * An avatar you already have saved (see [Characters](/en/modes/characters)).
    * A real photo (yours, a client's, etc.).
    * An image generated with [Image](/en/modes/image).
  </Step>

  <Step title="Take everything to Motion">
    In **Create → Motion**:

    * **CHARACTER** slot: the character image.
    * **MOTION VIDEO** slot: your motion clip.
    * Prompt: optional, the clip leads.
  </Step>

  <Step title="Generate">
    The result keeps the **original audio** from the reference clip.
    Useful when the audio was half the value of the video (talking
    gestures, laughs, etc.).

    Cost: 6 cr/s on Motion standard, 8 cr/s on Motion Pro.
  </Step>

  <Step title="Download and publish">
    Vertical 9:16 MP4, straight to TikTok / Reels / Shorts.
  </Step>
</Steps>

## Best practices

* **Face visible** in the motion clip: if you want the transfer to
  look natural, the camera has to see the face well.
* **Clear movement**, not too chaotic. A simple dance or a defined
  gesture works better than very fast motion.
* **Same framing**: if the reference clip is a medium shot, upload an
  image also at medium shot. Very different framing breaks the
  transfer.
* **Reuse the same avatar** across clips to build recognition (see
  [Build an AI influencer](/en/recipes/ai-influencer)).

## Common mistakes

* **Full-body image + close-up clip**: Motion doesn't invent a body.
  Match the image to the clip's framing.
* **Profile character**: loses detail. Front or 3/4 always.

<Note>
  Step-by-step walkthrough with examples at
  [zevor.ai/motion](https://zevor.ai/motion).
</Note>