HappyHorse 1.1 Image-to-Video API

Overview

HappyHorse 1.1 Image-to-Video is Alibaba's native multimodal AI video generation model (single-stream Transformer architecture) on iCreat: input a single starting image plus a text prompt — optionally with a negative prompt — to generate a short cinematic video with excellent temporal consistency, smooth and highly expressive character movement, and natively synchronized audio. Developers call ali/happyhorse-1-1/image-to-video through the async task queue, with 720P / 1080P output, five aspect ratios, and 4–15 second durations.

ali/happyhorse-1-1/image-to-video

Core Capabilities

Single-image animation. Transform one still image into a cinematic short video; the model keeps the subject and scene consistent while bringing the frame to life.

Excellent temporal consistency. The first frame anchors the video, so the scene stays stable frame to frame without drifting.

Smooth, expressive character motion. Character movements and micro-expressions animate fluidly from the still, avoiding stiff or robotic motion.

Text-driven motion and camera. The prompt describes how the scene unfolds — movement, camera push-ins, and atmosphere — rather than re-describing what is already visible in the image.

Negative prompt. negative_prompt describes content that should not appear in the video, giving fine control over what the model avoids.

Native audio-video synchronization. Built on a single-stream Transformer that encodes text, image, video, and audio in one representation space, the model generates synchronized audio (ambience, music, sound effects) together with the visuals.

Dual resolution. 720P for cost-effective iteration; 1080P for sharper final delivery.

Five aspect ratios. 1:1, 9:16, 16:9, 3:4, and 4:3 cover square feeds, vertical short video, and landscape formats.

4–15 second durations. Short lengths tailored for teasers, loops, and social clips.

Watermark control. watermark defaults to false; enable it when an AI-generated mark is required.

Async task design. Submit → poll → retrieve keeps the workflow batch-friendly, fault-tolerant, and easy to integrate.

Model Comparison

HappyHorse 1.1 Image-to-Video vs Seedance 2.5 Image-to-Video vs Kling V3 Omni

Input

The first-frame image and the text prompt drive generation. The negative prompt is optional.

Input Examples

Image-to-video

{
  "input": {
    "prompt": "A running kitten",
    "negative_prompt": "The kitten should not run in a straight line",
    "media": [
      {
        "type": "first_frame",
        "url": "https://example.com/cat.jpg"
      }
    ]
  },
  "parameters": {
    "resolution": "720P",
    "ratio": "1:1",
    "duration": 5,
    "watermark": false
  }
}

Without a negative prompt

{
  "input": {
    "prompt": "A kitten running across a grassy field",
    "media": [
      {
        "type": "first_frame",
        "url": "https://example.com/cat.jpg"
      }
    ]
  },
  "parameters": {
    "resolution": "1080P",
    "duration": 5,
    "watermark": false
  }
}

Parameters

input fields

parameters fields

Pricing

Total Cost = Unit Price × Output Video Duration

Quick Start

This API uses an API Key for authentication. Get your API Key from https://icreat.ai/hub/keys.

Submit a request:

curl --fail-with-body --connect-timeout 10 --max-time 60 \
  -X POST https://api.icreat.ai/v1/task/submit/ali/happyhorse-1-1/image-to-video \
  -H "Authorization: Bearer ${ICREAT_API_KEY}" \
  -H "Content-Type: application/json" \
  -d '{
    "input": {
      "prompt": "A running kitten",
      "negative_prompt": "The kitten should not run in a straight line",
      "media": [
        {
          "type": "first_frame",
          "url": "https://icreat.ai/xxx.jpg"
        }
      ]
    },
    "parameters": {
      "resolution": "720P",
      "duration": 5,
      "watermark": false
    }
  }'

The server will return:

{
  "task_id": "task-xxx"
}

Poll the task status:

curl --connect-timeout 10 --max-time 60 \
  -X POST "https://api.icreat.ai/v1/task/query-status" \
  -H "Authorization: Bearer ${ICREAT_API_KEY}" \
  -H "Content-Type: application/json" \
  -d '{
  "task_id": "task-xxx"
}'

When the status is SUCCEEDED, retrieve the result:

curl --connect-timeout 10 --max-time 60 \
  -X POST "https://api.icreat.ai/v1/task/get-result" \
  -H "Authorization: Bearer ${ICREAT_API_KEY}" \
  -H "Content-Type: application/json" \
  -d '{
  "task_id": "task-xxx"
}'

Output

submit → task_id → query-status (SUCCEEDED) → get-result

After submitting the request, poll query-status with the returned task_id until the status becomes SUCCEEDED, then call get-result with the same task_id to retrieve the generated video (with natively synchronized audio).

Use Cases

Bringing photos to life. Animate portraits, travel shots, and stills — a photo becomes a moving clip with subtle motion and native sound.

E-commerce product showcase. Use a product still as the first frame and drive the camera or product motion with a prompt — turn static listings into dynamic assets for feeds and detail pages.

Social short clips. 4–15 second teasers, loops, and clips sized for feeds and stories, with the aspect ratio matched to each platform.

Low-cost fast iteration. At $0.70 for a 720P 5-second clip, test multiple creative directions cheaply before investing in premium renders.

Limitations

Duration is limited to 4–15 seconds; longer narratives must be split across multiple requests and stitched in post.

Only 720P and 1080P resolutions are available.

The current interface accepts a single reference media type (first_frame); multi-image and video references (reference-to-video) are not supported on this endpoint.

resolution, ratio, and duration are required fields and must be included in every request.

FAQ

How do I get an API Key?

Create one at https://icreat.ai/hub/keys and pass it via the Authorization: Bearer header.

Is the first-frame image required?

The first-frame image is the core input of image-to-video: pass it via input.media with "type": "first_frame" and its URL. The schema marks media as optional, but for image-to-video you should always provide it; the text prompt is still required to describe the motion.

What does negative_prompt do?

Describes content that should not appear — e.g. "The kitten should not run in a straight line" — without touching the main prompt.

Is audio generated?

Yes. The model natively synchronizes audio with video. Audio is generated together with the visuals — no separate audio-generation request is required.

Is the watermark on by default?

No. watermark defaults to false; set it to true to add the AI-generated mark.

How is the cost calculated?

Total Cost = Unit Price × Output Video Duration. For example, a 10-second 1080P video costs 0.18 × 10 = $1.80.

720P or 1080P?

Use 720P for fast, low-cost iteration; use 1080P for final delivery that needs more visible detail.