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

# Video

> Create a video from text or one reference image, approve its credit quote, and download the finished MP4.

## Create a video with sound

Video is available through the DreamLayer Agent API. Describe a scene, or animate one uploaded image, then review a quote before authorizing generation. New requests default to sound on and a 10-second duration. You can explicitly request silence.

Use your API key and API credit balance. Check `GET /v1/video/capabilities` for current availability. Installation connections that allow image operations are not sufficient for Video. The guides below use REST and do not require a new CLI or MCP package release.

<CardGroup cols={2}>
  <Card title="Text to video" icon="video" href="/agent-api/text-to-video">
    Describe a scene, approve the quote, and start your first video.
  </Card>

  <Card title="Image to video" icon="image" href="/agent-api/image-to-video">
    Upload one first-frame reference and describe its movement.
  </Card>

  <Card title="Settings, sound, and pricing" icon="sliders" href="/agent-api/video-settings">
    Choose duration and audio, understand the quote, and set a spending ceiling.
  </Card>

  <Card title="Progress and recovery" icon="arrows-rotate" href="/agent-api/video-jobs">
    Poll, resume events, download, cancel eligible work, and recover safely.
  </Card>
</CardGroup>

## How Video differs from images

| | Ordinary images | Video |
| - | - | - |
| Start | Send a prompt to `/v1/execute` | Request `/v1/video/quotes`, then accept its `quote_id` at `/v1/execute` |
| Price | One credit per successful image | The quote's `customer_price.amount` in API credits |
| Submission response | JSON acknowledgement or an initial event stream | `202` JSON with the execution and `video_job` |
| Follow progress | Image execution states | Video-specific states, including `generating` and `uncertain` |
| Output | PNG | MP4, with sound on by default |
| Download | `/v1/image-jobs/{job_id}/asset` | `/v1/video/executions/{execution_id}/asset` |

Both use the same authenticated execution status and events endpoints. Video does not inherit a reference image or settings from an image conversation. Start a fresh quote with the intended prompt and source.

## Supported scope

* Text-to-video and one-reference-image image-to-video.
* Whole-second durations from 4 to 30 seconds, subject to quote validation; default 10 seconds.
* Native 720p output. Actual encoded dimensions can differ from an exact 720-pixel height. Read delivered metadata rather than assuming dimensions.
* Text video in `16:9`, `9:16`, or `1:1`. Image video uses the source-driven aspect profile.
* Generated audio or silent output. Audio is generated with the video, not uploaded separately.
* Authenticated download with 30-day output retention.

V1 does not accept multiple reference images, a source video to edit, an uploaded soundtrack, or arbitrary resolution requests. It does not automatically shorten or stitch a request outside its supported duration. Generated motion, speech, timing, and visual consistency are not guaranteed; review the result before using it in your project.

## Start here

Follow [text-to-video](/agent-api/text-to-video) for a complete quote-to-execution example. Use [image-to-video](/agent-api/image-to-video) when you have a reference. Read [limits](/agent-api/limits) before increasing polling or concurrency, and [Video recovery](/agent-api/video-jobs) before implementing retries.


This documentation is built and hosted on [Mintlify](https://mintlify.com), a developer documentation platform.