# RunwayML

**What is RunwayML?**

RunwayML is an AI-powered node in StackAI that enables users to generate high-quality images and videos using advanced generative models. It supports both text-to-image and image-to-video workflows, making it ideal for creative projects, content generation, and multimedia automation.

***

**How to use it?**

The RunwayML node offers two main actions:

1. **Generate Video from Image**
2. **Generate Image from Text**

Each action has its own set of inputs, configurations, and outputs. Below are detailed explanations and examples for each.

***

#### 1. Generate Video **from Image**

Create a video from an initial image, with optional text prompt guidance and customizable settings.

**Inputs:**

* **Image URL** (string, required):\
  The publicly accessible URL of the image to use as the first frame of the video.\
  \&#xNAN;*Example:* `"https://example.com/image.png"`
* **Prompt** (string, optional):\
  A text prompt to guide the video generation.\
  \&#xNAN;*Example:* `"A futuristic cityscape at sunset"`

**Configurations:**

* **Model** (select, optional, default: Gen4 Turbo):\
  The model variant for video generation.\
  \&#xNAN;*Options:* `"gen4_turbo"`
* **Duration** (select, optional, default: 5):\
  The length of the generated video in seconds.\
  \&#xNAN;*Options:* `5` or `10`
* **Aspect Ratio** (select, optional, default: 1280:720):\
  The output video’s aspect ratio.\
  \&#xNAN;*Options:*
  * 1280:720
  * 720:1280
  * 1104:832
  * 832:1104
  * 960:960
  * 1584:672
* **Seed** (number, optional):\
  Random seed for reproducible results (0-4294967295), using the same seed will produce the same image if all inputs and configurations are the same. \
  \&#xNAN;*Example:* `123456`
* **Watermark** (boolean, optional, default: true):\
  Whether to add a RunwayML watermark to the video.\
  \&#xNAN;*Options:* `true` or `false`

**Outputs:**

* **Video URL** (string):\
  The URL to download or view the generated video.
* **Task ID** (string):\
  The unique identifier for the video generation task.

**Example of Usage:**

```json
{
  "image_url": "https://example.com/image.png",
  "prompt": "A futuristic cityscape at sunset",
  "model": "gen4_turbo",
  "duration": 10,
  "aspect_ratio": "1280:720",
  "seed": 123456,
  "watermark": false
}
```

*Output:*

```json
{
  "video_url": "https://runwayml.com/generated/video123.mp4",
  "task_id": "task_abc123"
}
```

***

#### 2. Generate Image from Text

Create an image from a detailed text prompt, with options for model, resolution, and more.

**Inputs:**

* **Prompt** (string, required):\
  A detailed text description of the image to generate.\
  \&#xNAN;*Example:* `"A serene mountain landscape with a clear blue lake"`

**Configurations:**

* **Model** (select, optional, default: Gen4 Image):\
  The model to use for image generation.\
  \&#xNAN;*Options:* `"gen4_image"`
* **Resolution** (select, optional, default: 1024:1024):\
  The output image’s resolution/aspect ratio.\
  \&#xNAN;*Options include:*
  * 1920:1080 (16:9)
  * 1080:1920 (9:16)
  * 1024:1024 (Square)
  * 1360:768 (Landscape)
  * 1080:1080 (Square)
  * 1168:880 (4:3)
  * 1440:1080 (4:3)
  * 1080:1440 (3:4)
  * 1808:768 (Wide)
  * 2112:912 (Ultra Wide)
  * 1280:720 (HD)
  * 720:1280 (Portrait HD)
  * 720:720 (Square HD)
  * 960:720 (4:3 HD)
  * 720:960 (3:4 HD)
  * 1680:720 (Cinematic)
* **Seed** (number, optional):\
  Random seed for reproducible results (0-4294967295), using the same seed will produce the same image if all inputs and configurations are the same. \
  \&#xNAN;*Example:* `78910`
* **Public Figure Threshold** (select, optional, default: auto):\
  Content moderation strictness for public figures.\
  \&#xNAN;*Options:* `"auto"`, `"low"`
* **Reference Images** (string array, optional):\
  Array of image URLs to use as references for generation. Can help augment the style of the image created.\
  \&#xNAN;*Example:* `["https://example.com/ref1.jpg", "https://example.com/ref2.jpg"]`

**Outputs:**

* **Image URL** (string):\
  The URL to download or view the generated image.
* **Task ID** (string):\
  The unique identifier for the image generation task.

**Example of Usage:**

```json
{
  "prompt": "A serene mountain landscape with a clear blue lake",
  "model": "gen4_image",
  "resolution": "1920:1080",
  "seed": 78910,
  "public_figure_threshold": "auto",
  "reference_images": ["https://example.com/ref1.jpg"]
}
```

*Output:*

```json
{
  "image_url": "https://runwayml.com/generated/image456.png",
  "task_id": "task_def456"
}
```

***

**Summary Table**

| Action         | Required Inputs | Optional Configurations                                               | Outputs (Required)   |
| -------------- | --------------- | --------------------------------------------------------------------- | -------------------- |
| Generate Video | image\_url      | prompt, model, duration, aspect\_ratio, seed, watermark               | video\_url, task\_id |
| Generate Image | prompt          | model, resolution, seed, public\_figure\_threshold, reference\_images | image\_url, task\_id |

***

**Advanced Settings**

* Retry on Failure: Enable retrying when the node execution fails
* Fallback Branch: Create a separate branch that executes when this node fails, allowing you to handle errors gracefullyCreate a separate branch that executes when this node fails, allowing you to handle errors gracefully