Aliyun Happyhorse Videoedit

HappyHorse 1.0 Video Editing

Validation

bashCopy code
mkdir -p output/aliyun-happyhorse-videoedit
python -m py_compile skills/ai/video/aliyun-happyhorse-videoedit/scripts/edit_happyhorse.py && echo "py_compile_ok" > output/aliyun-happyhorse-videoedit/validate.txt

Pass criteria: command exits 0 and output/aliyun-happyhorse-videoedit/validate.txt is generated.

Output And Evidence

• Save task IDs, polling responses, and final video URLs to output/aliyun-happyhorse-videoedit/.
• Keep at least one end-to-end run log for troubleshooting.

Prerequisites

• Install dependencies (recommended in a venv):

bashCopy code
python3 -m venv .venv
. .venv/bin/activate
python -m pip install requests

• Set DASHSCOPE_API_KEY in your environment, or add dashscope_api_key to ~/.alibabacloud/credentials.

Critical model names

• happyhorse-1.0-video-edit — instruction-based video editing with optional reference images and audio retention control

Capabilities

API endpoint (async only)

Copy code
POST https://dashscope.aliyuncs.com/api/v1/services/aigc/video-generation/video-synthesis

Required headers:

• Authorization: Bearer $DASHSCOPE_API_KEY
• Content-Type: application/json
• X-DashScope-Async: enable

Singapore endpoint: replace dashscope.aliyuncs.com with dashscope-intl.aliyuncs.com.

Polling endpoint: GET https://dashscope.aliyuncs.com/api/v1/tasks/{task_id} — recommended interval 15s.

Normalized interface

Request

• model (string, required) — fixed happyhorse-1.0-video-edit
• input.prompt (string, required) — up to 5000 non-CJK / 2500 CJK characters describing the edit
• input.media (array, required) — exactly 1 video element, plus 0-5 reference_image elements:
  • type: video (required, exactly 1) | reference_image (optional, 0-5)
  • url: public HTTP/HTTPS URL
• parameters.resolution (string, optional) — 720P or 1080P (default: 1080P)
• parameters.audio_setting (string, optional) — auto (default, model decides) or origin (keep input audio)
• parameters.watermark (boolean, optional) — bottom-right "Happy Horse" watermark (default: true)
• parameters.seed (integer, optional) — range [0, 2147483647]

Media input limits

Input video (type=video):

• Formats: MP4, MOV (H.264 encoding recommended)
• Duration: 3-60 seconds (output is capped at 15s; videos >15s are truncated to the first 15s)
• Resolution: long side ≤ 2160 px, short side ≥ 320 px
• Aspect ratio: 1:2.5 ~ 2.5:1
• Frame rate: > 8 fps
• Max size: 100 MB

Reference image (type=reference_image):

• Formats: JPEG, JPG, PNG, WEBP
• Resolution: width and height ≥ 300 pixels
• Aspect ratio: 1:2.5 ~ 2.5:1
• Max size: 10 MB

Output duration rule

• Input ≤ 15s → output duration = input duration.
• Input > 15s → input is truncated to the first 15s; output ≤ 15s.

Response (task creation)

• output.task_id (string) — valid 24 hours
• output.task_status (string) — PENDING | RUNNING | SUCCEEDED | FAILED | CANCELED | UNKNOWN
• request_id (string)

Response (task result, on SUCCEEDED)

• output.video_url (string) — edited MP4 (H.264) URL, valid 24 hours
• output.orig_prompt (string)
• output.submit_time / output.scheduled_time / output.end_time (string)
• usage.duration (float) — billable duration in seconds
• usage.input_video_duration (float)
• usage.output_video_duration (float)
• usage.SR (integer) — output resolution tier
• usage.video_count (integer) — fixed 1

Quick start (Python + HTTP)

pythonCopy code
import os
import time
import requests

API_KEY = os.getenv("DASHSCOPE_API_KEY")
BASE_URL = "https://dashscope.aliyuncs.com/api/v1"


def create_videoedit_task(req: dict) -> str:
    """Create a video-edit task and return task_id."""
    media = [{"type": "video", "url": req["video_url"]}]
    for url in req.get("reference_images", []):
        media.append({"type": "reference_image", "url": url})
    if len(media) - 1 > 5:
        raise ValueError("At most 5 reference images")

    payload = {
        "model": "happyhorse-1.0-video-edit",
        "input": {"prompt": req["prompt"], "media": media},
        "parameters": {
            "resolution": req.get("resolution", "1080P"),
            "watermark": req.get("watermark", True),
        },
    }
    if req.get("audio_setting"):
        payload["parameters"]["audio_setting"] = req["audio_setting"]
    if req.get("seed") is not None:
        payload["parameters"]["seed"] = req["seed"]

    resp = requests.post(
        f"{BASE_URL}/services/aigc/video-generation/video-synthesis",
        headers={
            "Authorization": f"Bearer {API_KEY}",
            "Content-Type": "application/json",
            "X-DashScope-Async": "enable",
        },
        json=payload,
    )
    resp.raise_for_status()
    return resp.json()["output"]["task_id"]


def poll_task(task_id: str, interval: int = 15) -> dict:
    while True:
        resp = requests.get(
            f"{BASE_URL}/tasks/{task_id}",
            headers={"Authorization": f"Bearer {API_KEY}"},
        )
        resp.raise_for_status()
        data = resp.json()
        if data["output"]["task_status"] in ("SUCCEEDED", "FAILED", "CANCELED"):
            return data
        time.sleep(interval)

Usage examples

pythonCopy code
# Style transfer — instruction only, no reference image
task_id = create_videoedit_task({
    "video_url": "https://example.com/input.mp4",
    "prompt": "将整个画面转换为水墨画风格",
    "resolution": "720P",
})

# Local replacement with a reference image, keep original audio
task_id = create_videoedit_task({
    "video_url": "https://example.com/character.mp4",
    "reference_images": ["https://example.com/striped-sweater.webp"],
    "prompt": "让视频中的马头人身角色穿上图片中的条纹毛衣",
    "audio_setting": "origin",
})

Error handling

Output location

• Default output: output/aliyun-happyhorse-videoedit/videos/
• Override base dir with OUTPUT_DIR.

Anti-patterns

• Do not use any model ID other than happyhorse-1.0-video-edit.
• Do not call this API synchronously — async header is required.
• Do not pass more than 1 video, more than 5 reference images, or any first_frame / last_frame / driving_audio.
• Do not pass ratio or duration — output ratio follows the input video and duration follows the truncation rule.
• Video URLs expire after 24 hours; download and persist immediately.
• Do not use this skill for generation from scratch — use aliyun-happyhorse-t2v, aliyun-happyhorse-i2v, or aliyun-happyhorse-r2v instead.

Workflow

1. Confirm intent: style transfer vs. instruction edit, and whether reference images are needed.
2. Validate the input video meets format / duration / resolution / fps / size limits.
3. Build the media array with exactly 1 video plus 0-5 reference_image entries.
4. Create async task and poll /tasks/{task_id} every ~15s; download output.video_url before 24-hour expiration.

References

• See references/api_reference.md for full HTTP API details.
• See references/sources.md for source links.