Kandinsky 6.0 Pro Image to Video API

POST Base URL: https://api.hiapi.ai /v1/tasks

Copy Page

Image, video, and audio models are called through theUnified Async API POST /v1/tasks endpoint; only the input fields differ (see input parameters below).

Model summary

Model	kandinsky-6.0-pro/image-to-video
Capability	First-frame image to video with synchronized audio
Resolution	480p (16:9 delivered at 864×480)
Length	Fixed 5 seconds

Animate a first frame into a 5-second 480p video with synchronized audio and automatic aspect matching. Billed per video.

Production guidance

Calling tips

• The first frame sets subject and composition; the prompt adds only motion, camera and sound.
• aspect_ratio=auto picks the closest supported frame and center-crops the image; set 16:9, 9:16, 1:1, 4:3 or 3:4 explicitly for exact control.
• Use a publicly reachable JPG, PNG or WebP image, and send the URL returned by your upload.
• Use callback.url for notifications; result URLs expire, so copy clips to your own storage.

Best suited for

Animate product shots

Add an orbit, push-in and sound to an existing product photo while the product keeps its shape.

image_urlprompt

Posters and illustrations

Turn a still poster or illustration into a 5-second motion asset.

image_urlaspect_ratio

Add motion to real scenes

Bring rain, mist or passing people into a landscape, street or architecture photo.

promptseed

Request parameters

</> Generate example code

model string required

Model ID.

example kandinsky-6.0-pro/image-to-video

input object required

Image-to-video parameters.

prompt string required

Describe the motion, camera and sound to add; no need to restate the image. Up to 6,000 characters; English or Russian works best.

image_url string required

Image used as the first frame; JPG, PNG or WebP. With aspect_ratio=auto the closest supported frame is chosen and the image is center-cropped to it.

negative_prompt string optional

Content or artifacts to avoid, up to 2,000 characters.

aspect_ratio enum optional

Output frame. auto picks the supported frame closest to the first frame and center-crops to it; the other values match text-to-video.

default auto enum: auto16:99:161:14:33:4

seed integer optional

Random seed for reproducing a generation. Omit for a random seed.

guidance_scale number optional

How closely the result follows the prompt; a number from 1 to 10, default 5. Higher values follow the text more strictly.

default 5

callback object optional

Sends a notification when the task succeeds or fails; read the task object from data in query responses and directly from the callback body.

url string required

HTTPS URL that receives task success or failure notifications.

example https://your-domain.com/hiapi/callback

when enum optional

Triggers when the task succeeds or fails.

default final enum: final

Input requirements

• prompt and image_url are both required; prompt is up to 6,000 characters and negative_prompt up to 2,000.
• The image is the first frame and must be JPG, PNG or WebP; aspect_ratio=auto picks the closest supported frame and center-crops to it.
• aspect_ratio only accepts auto, 16:9, 9:16, 1:1, 4:3 or 3:4; a 16:9 clip is delivered at 864×480.
• Length is fixed at 5 seconds, resolution at 480p and audio is always on; output is a 24fps MP4.
• guidance_scale accepts numbers from 1 to 10; seed is an integer.

Example requests

Rainy phone booth

A 16:9 street photo as the first frame, auto frame, fixed 5 seconds at 480p with synchronized audio.

Request body⧉

{
  "model": "kandinsky-6.0-pro/image-to-video",
  "input": {
    "prompt": "Rain intensifies on the empty street, neon reflections ripple in the puddles, steam rises from a grate beside the booth, the camera slowly pushes in, steady rain and distant traffic",
    "image_url": "https://your-domain.com/first-frame.jpg",
    "aspect_ratio": "auto"
  }
}

Vertical portrait with motion

A vertical first frame with 9:16 output for short-video feeds.

Request body⧉

{
  "model": "kandinsky-6.0-pro/image-to-video",
  "input": {
    "prompt": "The subject turns slightly toward the window, hair moves in the breeze, soft city ambience",
    "image_url": "https://your-domain.com/portrait.jpg",
    "aspect_ratio": "9:16"
  }
}

Getting the result

1. Read data.taskId from the create-task response.
2. Wait for the callback or query GET /v1/tasks/:id; polling puts the task under data.
3. When status=success, read the video URL from output type=video; a callback reads output directly.

FAQ

What are the first-frame image requirements?

A publicly reachable JPG, PNG or WebP URL. The image sets subject and composition; the output frame comes from aspect_ratio.

How does auto handle the image?

It picks the supported frame closest to the image and center-crops to it. Prepare the image in the target ratio if cropping is not acceptable.

What should the prompt describe?

Only the motion, camera and sound to add, not what the image already shows. English and Russian are natively supported.

How long can the clip be?

Fixed at 5 seconds and 480p; duration, resolution, step count and upscaling are not exposed.

Can I send an image without a prompt?

No. prompt is required and directs motion, camera and sound.

Can I reproduce a result?

Pass the same image, seed and parameters to reproduce it as closely as possible; small differences can remain between runs.

How is it billed?

Per video: each successful generation is charged one fixed tier price. See the pricing page for the current amount.

Do result links expire?

Yes. Download or copy clips to your own storage if you need to keep them.

Next steps

• Text to Video
• View live pricing