fal-ai/kling-video/v3/pro/image-to-video

Run Kling 3.0 Image To Video Pro API on fal

Kling 3.0 Pro image-to-video on fal.ai. Cinematic visuals, fluid motion, native audio generation, and custom element support.

Features

• Videos up to 15 seconds
• Multi-prompt support for multi-scene narrative control
• Custom element injection via the `elements` parameter (reference characters/objects)
• Native audio with multiple speakers and language support
• Strong subject and text consistency
• Aspect ratio is determined by the start image, not a parameter

Pricing

Quick Start

Install

bash
npm install --save @fal-ai/client

bash
export FAL_KEY="YOUR_API_KEY"

Submit a request

javascript
import { fal } from "@fal-ai/client";

const result = await fal.subscribe("fal-ai/kling-video/v3/pro/image-to-video", {
  input: {
    start_image_url: "https://example.com/your-image.png",
    prompt: "Slow cinematic push-in. Golden light. No people.",
    duration: "10",
    generate_audio: true,
  },
  logs: true,
  onQueueUpdate: (update) => {
    if (update.status === "IN_PROGRESS") {
      update.logs.map((log) => log.message).forEach(console.log);
    }
  },
});

console.log(result.data.video.url);

Input Parameters

Custom Elements

Inject a reference character or object into the video using the `elements` array. Reference them in your prompt as `@Element1`, `@Element2`, etc.

Each element can be either an image set (frontal + optional reference images) or a video:

json
{
  "elements": [
    {
      "frontal_image_url": "https://example.com/character-front.png",
      "reference_image_urls": ["https://example.com/character-side.png"]
    }
  ]
}

Note: Voice binding is only supported for video elements, not image elements. Attempting voice binding with an image element returns an error.

Multi-Prompt (Multi-Shot)

Divide the video into multiple shots, each with its own prompt and duration:

javascript
{
  multi_prompt: [
    { prompt: "Wide establishing shot of the temple at dawn.", duration: "5" },
    { prompt: "Close-up on the warrior's face. Wind in his hair.", duration: "5" },
  ],
  shot_type: "customize",
  start_image_url: "https://example.com/scene.png",
  duration: "10",
}

Output

json
{
  "video": {
    "url": "https://storage.googleapis.com/...",
    "content_type": "video/mp4",
    "file_name": "out.mp4",
    "file_size": 8431922
  }
}

Infrastructure

• Endpoint alias for concurrency tracking: `fal-ai/kling-video-v3`
• Default concurrency limit: 1 per user (overrides available on request)
• Playground variant: `fal-ai/kling-video/v3/pro/image-to-video/playground`
• Queue-based: for long jobs, use `fal.queue.submit` + webhook instead of blocking

Known Limitations

• Aspect ratio is inferred from the start image. The `aspect_ratio` field in the UI is ignored by the model.
• Voice binding is only supported for video elements, not image elements.
• Audio language support: Chinese and English natively. Other languages are auto-translated to English.

Related Endpoints