> Index of Phersa pages for AI agents: https://phersa.com/llms.txt

# AI Video Prompt Generator

Write an AI video prompt that turns an image of your AI persona into a clip: what happens, the line they say, the camera move, the light, the mood and the sound. It is for anyone making talking clips of an AI influencer with Gemini Omni 1.1 Flash or Grok Imagine Video 1.5, and it checks that your line fits the clip.

## How to use the AI Video Prompt Generator

1. Say what is in the start image, pick what happens, and type the line your persona says, if any.

2. Choose the camera move, light, mood, sound, model and length. Randomize tries new picks and keeps your line.

3. Copy the prompt and attach the start image in your video tool. Set the model, shape and length shown under the prompt.

## The parts of a video prompt

A clip prompt starts from an image and adds what the image can’t show: movement, sound and time.

- **Start image.** The attached image is the first frame. One line says so, and says the person keeps their face, hair and clothes.
- **Action.** One thing that happens, from start to end: lifts the mug, sips, smiles. A few seconds hold one or two moves.
- **Spoken line.** The exact words in quotes. Google’s guide for its Veo video model says to use quotes for specific speech. The prompt adds that the persona says only these words.
- **Camera.** Plain camera words: still or “locked off”, “push in”, a “natural smartphone zoom”, “one continuous shot”. Google’s Gemini Omni guide uses these terms.
- **Light and mood.** Keep the image’s light, or name a change. The mood is two words.
- **Sound.** Say what you want to hear: the voice, the room, soft music. Google’s Omni docs say the prompt can describe the audio track.

## Keeping the face from the start image

The video model takes the face from the first frame. A clip looks like your persona only when that image does, so make the image first and check it.

- **Start from the persona’s image.** Make the clip’s image with the persona’s reference photo attached, then animate that image.
- **Say what stays.** “Keeps the exact face, hair and clothes from the image” is in every prompt here.
- **Keep the face in view.** Small head movements, a turn toward the lens, a smile. Short clips with one action stay closer to the image.
- **One line per clip.** Split a longer message across clips, and join them in your editor.

## Clip length and shape by model

| Model | Length | Shape |
| --- | --- | --- |
| Gemini Omni 1.1 Flash | 3 to 10 seconds, with sound | 9:16 or 16:9 |
| Grok Imagine Video 1.5 | Up to 15 seconds | From one image, the image’s shape |

Some providers offer fewer lengths for the same model, such as 4, 6, 8 or 10 seconds. At a relaxed pace people say about 150 words a minute, roughly 2.5 words a second, so a 6 second clip fits a line of about 12 words with a breath at the end.

Make each clip’s image from your AI persona’s photo, then its video from that image, on your own API keys with Gemini Omni or Grok Imagine. [How making videos works →](https://phersa.com/docs/create)

## AI Video Prompt Generator questions

### How do I make an AI influencer talk?

Write the exact line in quotes in the video prompt, and use a model that makes sound with the video, such as Gemini Omni or Grok Imagine Video 1.5. Start from a sharp image of the persona facing the camera. Keep the line short enough for the clip; the tool warns you when it isn’t.

### How long can the spoken line be?

About 2.5 words a second, a relaxed talking pace, with half a second free at the end. A 6 second clip fits about 12 words. The tool counts your words and warns you when the line is longer than the clip.

### Why does my persona’s face change in the video?

The model builds every frame from the first one, so small changes add up. Start from a sharp image where the face is clear, keep the action small and the clip short. If it still drifts, make the clip again from the same image.

[Consistent Character Prompt Generator →](https://phersa.com/tools/consistent-character-prompt-generator)

### Should the start image be the same shape as the video?

Yes. For a Reel or a Story, make the start image 9:16, the shape Meta lists for both. The first frame sets what the clip shows, and Grok Imagine takes the video’s shape from the image.

### Do these models make sound?

Yes. Gemini Omni makes the audio with the video, and Grok Imagine Video 1.5 makes sound too. Describe what you want to hear in the prompt; the Sound pick does that for you.

### Do I have to label AI videos on Instagram?

Yes, when the video looks real. Meta requires its AI label when you post a photorealistic video or realistic-sounding audio that was digitally created or altered, and it may apply penalties if you don’t.

## More free tools

- [AI Image Prompt Generator — A photo prompt for your AI persona: setting, outfit, pose, light, camera and framing, shaped for your model.](https://phersa.com/tools/ai-image-prompt-generator)
- [Consistent Character Prompt Generator — One identity block, many shots: image prompts that repeat the same face, hair and look word for word.](https://phersa.com/tools/consistent-character-prompt-generator)
- [Shot List Generator — Plan a short vertical video clip by clip, timed to a target length, and export it as CSV or JSON.](https://phersa.com/tools/shot-list-generator)

[See all free tools for AI persona creators →](https://phersa.com/tools)

Source: https://phersa.com/tools/ai-video-prompt-generator