Omni 视频生成¶

Google Omni 模型，支持文生视频、图生视频和视频修改三种模式，异步提交 + 轮询查询。

快速开始¶

# 1. 提交任务（图生视频）
curl -X POST https://agent.tweapi.com/v1/videos \
  -H "Authorization: Bearer $API_KEY" \
  -H "Content-Type: application/json" \
  -d '{
    "model": "omni_flash-10s",
    "prompt": "根据参考图生成视频",
    "size": "1280x720",
    "images": [
      "https://example.com/img1.jpg",
      "https://example.com/img2.jpg",
      "https://example.com/img3.jpg"
    ]
  }'

# 2. 轮询结果
curl https://agent.tweapi.com/v1/videos/{task_id} \
  -H "Authorization: Bearer $API_KEY"

功能模式¶

模式	说明	传参
文生视频	纯提示词生成	不传 `images` / `input_reference[]`
图生视频	参考图生成（最多 7 张）	传图片 URL
视频修改	上传视频 + 参考图进行修改	传视频 URL + 图片 URL

可用模型¶

模型	时长	分辨率
`omni_flash-10s`	10s	720p

通用说明¶

端点¶

方法	路径	说明
`POST`	`/v1/videos`	提交视频生成任务
`GET`	`/v1/videos/{task_id}`	查询任务状态

请求头¶

参数	值	说明
`Authorization`	`Bearer YOUR_API_KEY`	必填
`Content-Type`	`application/json` 或 `multipart/form-data`	POST 时必填

请求参数¶

参数	类型	必填	说明
`model`	string	是	模型名称：`omni_flash-10s`
`prompt`	string	是	文本提示词
`size`	string	否	视频尺寸，如 `1280x720`（横屏）、`720x1280`（竖屏）
`images`	string[]	否	图片/视频 URL 数组，最多 7 项（JSON 方式使用此字段）
`input_reference[]`	可重复	否	图片/视频 URL，最多 7 项（表单方式使用此字段，同一字段名可重复多次）

JSON 与表单二选一即可。文生视频时不传图片/视频字段。

响应参数¶

参数	类型	说明
`id`	string	任务 ID
`status`	string	`queued` / `processing` / `completed` / `failed`
`progress`	number	进度 0~100
`video_url`	string	视频直链（仅 `completed` 时返回）
`error`	object	错误信息（仅 `failed` 时返回），含 `message`、`code`

状态流转¶

queued → processing → completed
                   ↘ failed

提交任务¶

JSON 方式 — 图生视频（多图）¶

curl -X POST "https://agent.tweapi.com/v1/videos" \
  -H "Authorization: Bearer YOUR_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{
    "model": "omni_flash-10s",
    "prompt": "根据参考图生成视频",
    "size": "1280x720",
    "images": [
      "https://example.com/img1.jpg",
      "https://example.com/img2.jpg",
      "https://example.com/img3.jpg"
    ]
  }'

JSON 方式 — 视频修改（视频 URL + 图片 URL）¶

curl -X POST "https://agent.tweapi.com/v1/videos" \
  -H "Authorization: Bearer YOUR_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{
    "model": "omni_flash-10s",
    "prompt": "将视频中的场景改为参考图的风格",
    "size": "1280x720",
    "images": [
      "https://example.com/input.mp4",
      "https://example.com/reference.png"
    ]
  }'

表单方式 — 图生视频（多图）¶

curl -X POST "https://agent.tweapi.com/v1/videos" \
  -H "Authorization: Bearer YOUR_API_KEY" \
  -F "model=omni_flash-10s" \
  -F "prompt=根据参考图生成视频" \
  -F "size=1280x720" \
  -F "input_reference[]=https://example.com/img1.jpg" \
  -F "input_reference[]=https://example.com/img2.jpg" \
  -F "input_reference[]=https://example.com/img3.jpg"

表单方式 — 视频修改（视频 URL + 图片 URL）¶

curl -X POST "https://agent.tweapi.com/v1/videos" \
  -H "Authorization: Bearer YOUR_API_KEY" \
  -F "model=omni_flash-10s" \
  -F "prompt=将视频中的场景改为参考图的风格" \
  -F "size=1280x720" \
  -F "input_reference[]=https://example.com/input.mp4" \
  -F "input_reference[]=https://example.com/reference.png"

提交响应示例¶

{
  "id": "task_xxxxxxxxxxxx",
  "status": "queued",
  "progress": 0
}

查询任务¶

GET /v1/videos/{task_id}

curl "https://agent.tweapi.com/v1/videos/task_xxxxxxxxxxxx" \
  -H "Authorization: Bearer YOUR_API_KEY"

进行中响应¶

{
  "id": "task_xxxxxxxxxxxx",
  "status": "processing",
  "progress": 45
}

完成响应¶

{
  "id": "task_xxxxxxxxxxxx",
  "status": "completed",
  "progress": 100,
  "video_url": "https://img.12ai.org/videos/task_xxxxxxxxxxxx.mp4"
}

失败响应¶

{
  "id": "task_xxxxxxxxxxxx",
  "status": "failed",
  "error": {
    "message": "内容安全策略拦截",
    "code": "content_filter"
  }
}

完整调用示例¶

Python — 图生视频（JSON）Python — 视频修改（JSON）Python — 视频修改（表单）纯 cURL 流程

import requests
import time

API_BASE = "https://agent.tweapi.com"
API_KEY = "YOUR_API_KEY"
HEADERS = {
    "Authorization": f"Bearer {API_KEY}",
    "Content-Type": "application/json",
}

# 1. 提交任务
resp = requests.post(
    f"{API_BASE}/v1/videos",
    headers=HEADERS,
    json={
        "model": "omni_flash-10s",
        "prompt": "根据参考图生成视频",
        "size": "1280x720",
        "images": [
            "https://example.com/img1.jpg",
            "https://example.com/img2.jpg",
            "https://example.com/img3.jpg",
        ],
    },
)
task = resp.json()
task_id = task["id"]
print(f"任务已创建: {task_id}")

# 2. 轮询状态
while True:
    resp = requests.get(
        f"{API_BASE}/v1/videos/{task_id}",
        headers={"Authorization": f"Bearer {API_KEY}"},
    )
    data = resp.json()
    status = data["status"]
    progress = data.get("progress", 0)
    print(f"状态: {status}, 进度: {progress}%")

    if status == "completed":
        print(f"生成成功！视频地址: {data['video_url']}")
        break
    elif status == "failed":
        print(f"生成失败: {data['error']['message']}")
        break

    time.sleep(5)

import requests
import time

API_BASE = "https://agent.tweapi.com"
API_KEY = "YOUR_API_KEY"

# 1. 提交视频修改任务
resp = requests.post(
    f"{API_BASE}/v1/videos",
    headers={
        "Authorization": f"Bearer {API_KEY}",
        "Content-Type": "application/json",
    },
    json={
        "model": "omni_flash-10s",
        "prompt": "将视频中的场景改为参考图的风格",
        "size": "1280x720",
        "images": [
            "https://example.com/input.mp4",
            "https://example.com/reference.png",
        ],
    },
)
task_id = resp.json()["id"]
print(f"任务已创建: {task_id}")

# 2. 轮询状态
while True:
    resp = requests.get(
        f"{API_BASE}/v1/videos/{task_id}",
        headers={"Authorization": f"Bearer {API_KEY}"},
    )
    data = resp.json()
    if data["status"] == "completed":
        print("视频地址:", data["video_url"])
        break
    elif data["status"] == "failed":
        print("失败:", data["error"]["message"])
        break
    time.sleep(5)

import requests
import time

API_BASE = "https://agent.tweapi.com"
API_KEY = "YOUR_API_KEY"

# 1. 提交视频修改任务（multipart/form-data）
resp = requests.post(
    f"{API_BASE}/v1/videos",
    headers={"Authorization": f"Bearer {API_KEY}"},
    files=[
        ("model", (None, "omni_flash-10s")),
        ("prompt", (None, "将视频中的场景改为参考图的风格")),
        ("size", (None, "1280x720")),
        ("input_reference[]", (None, "https://example.com/input.mp4")),
        ("input_reference[]", (None, "https://example.com/reference.png")),
    ],
)
task_id = resp.json()["id"]
print(f"任务已创建: {task_id}")

# 2. 轮询状态
while True:
    resp = requests.get(
        f"{API_BASE}/v1/videos/{task_id}",
        headers={"Authorization": f"Bearer {API_KEY}"},
    )
    data = resp.json()
    if data["status"] == "completed":
        print("视频地址:", data["video_url"])
        break
    elif data["status"] == "failed":
        print("失败:", data["error"]["message"])
        break
    time.sleep(5)

# 1. 创建任务
ID=$(curl -s -X POST https://agent.tweapi.com/v1/videos \
  -H "Authorization: Bearer $API_KEY" \
  -H "Content-Type: application/json" \
  -d '{
    "model": "omni_flash-10s",
    "prompt": "根据参考图生成视频",
    "size": "1280x720",
    "images": [
      "https://example.com/img1.jpg",
      "https://example.com/img2.jpg",
      "https://example.com/img3.jpg"
    ]
  }' | jq -r '.id')
echo "task id: $ID"

# 2. 轮询状态
while true; do
  RESP=$(curl -s "https://agent.tweapi.com/v1/videos/$ID" \
           -H "Authorization: Bearer $API_KEY")
  STATUS=$(echo "$RESP" | jq -r '.status')
  echo "$STATUS"

  if [ "$STATUS" = "completed" ]; then
    echo "视频地址: $(echo "$RESP" | jq -r '.video_url')"
    break
  elif [ "$STATUS" = "failed" ]; then
    echo "失败: $(echo "$RESP" | jq -r '.error.message')"
    exit 1
  fi

  sleep 5
done

注意事项¶

图片/视频上限共 7 项，按传入顺序对应
URL 须为**公网可访问的直链**，不要传网页地址或需登录的链接
JSON 与表单**二选一**：JSON 方式使用 images 数组，表单方式使用 input_reference[] 重复字段
文生视频时不传 images / input_reference[] 即可
视频修改：将视频 URL 混入 images / input_reference[] 中即可，系统自动识别
异步接口，提交后需**轮询查询结果**，建议每 5 秒轮询一次