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

# Create image-to-video job

## Hailuo image-to-video

Upload a start frame and animate it with text + camera commands.

***

## Quick flow

```
1. Submit job → get task_id
2. Poll status → wait until status is completed for a public URL
```

### Basics

* **Endpoint**: `POST /v1/video/generations`
* **Auth**: `Bearer Token` (`Authorization: Bearer {{YOUR_API_KEY}}`)
* **Content-Type**: `application/json`

***

### Request body

| Field                      | Type      | Required | Description                                                                   |
| :------------------------- | :-------- | :------- | :---------------------------------------------------------------------------- |
| **model**                  | `string`  | **Yes**  | e.g. `MiniMax-Hailuo-2.3`, `MiniMax-Hailuo-2.3-Fast`, `I2V-01-Director`.      |
| **prompt**                 | `string`  | **Yes**  | Motion description, max 2000 chars; optional bracket camera tags (see below). |
| **duration**               | `integer` | **Yes**  | `6` or `10` (model-dependent).                                                |
| **metadata**               | `object`  | **Yes**  | Extensions.                                                                   |
| └─ **first\_frame\_image** | `string`  | **Yes**  | Start frame: public URL or base64 data URL.                                   |
| └─ **resolution**          | `string`  | No       | `512P`, `768P`, `1080P`.                                                      |
| └─ **prompt\_optimizer**   | `boolean` | No       | Auto refine prompt (default true).                                            |
| └─ **callback\_url**       | `string`  | No       | Status webhook.                                                               |
| └─ **aigc\_watermark**     | `boolean` | No       | Watermark (default false).                                                    |

***

### Image rules

* **Formats**: JPG, JPEG, PNG, WebP
* **Size**: \< 20MB
* **Dimensions**: short edge > 300px; aspect between 2:5 and 5:2

***

### Camera commands (sample)

* **Dolly**: `[dolly in]`, `[dolly out]`
* **Truck / pedestal**: `[truck left]`, `[truck right]`, `[pedestal up]`, `[pedestal down]`
* **Pan / tilt**: `[pan left]`, `[pan right]`, `[tilt up]`, `[tilt down]`
* **Zoom**: `[zoom in]`, `[zoom out]`

> **Compatibility:** Production may require **Chinese** bracket tags per the vendor. If needed, copy those literals from the Chinese documentation.

***

### Example

```json theme={null}
{
  "model": "MiniMax-Hailuo-2.3",
  "prompt": "Contemporary dance, the people in the picture are performing, [dolly in]",
  "duration": 6,
  "metadata": {
    "first_frame_image": "https://example.com/start_frame.png",
    "resolution": "1080P",
    "prompt_optimizer": true,
    "aigc_watermark": false
  }
}
```

***

### Response

**200 OK**

| Field        | Type     | Description                  |
| :----------- | :------- | :--------------------------- |
| **task\_id** | `string` | Id for status polling.       |
| **status**   | `string` | e.g. `queued`, `processing`. |

***

### Notes

1. **Align prompt with image**: Describe motion that fits the frame; mismatched action vs. static scene may look poor.
2. **Speed**: `MiniMax-Hailuo-2.3-Fast` for faster turnaround.
3. **10s + `MiniMax-Hailuo-02`**: max `768P`.