OmniHuman v2 API — 如何接入并付费

什么是 OmniHuman v2

OmniHuman v2 — 新一代 OmniHuman（October 2026）以及推荐版本：根据照片和音轨生成逼真的人物动画——嘴唇、面部表情、手势和身体动作都会跟随您的音频。需要 image_url 和 audio_url（wav 或 mp3，4 秒至 5 分钟）；片段时长与音频相同。可选提示词（英文）可控制场景和镜头。价格低于 OmniHuman 1.5。按成品视频的每秒计价。

请求示例

在 Bearer 请求头中传递密钥。API 会返回 jobId 用于轮询状态。

curl -sS https://getvibeapi.com/api/media/generate \
  -H "Authorization: Bearer $GETVIBE_API_KEY" \
  -H "Content-Type: application/json" \
  --data-binary '{"model":"omnihuman-v2:flow","prompt":"a beautiful landscape","opts":{"image_url":"https://example.com/image_url.png","audio_url":"https://example.com/audio_url.mp3"}}'

JavaScript

const response = await fetch('https://getvibeapi.com/api/media/generate', {
  method: 'POST',
  headers: {
    Authorization: `Bearer ${process.env.GETVIBE_API_KEY}`,
    'Content-Type': 'application/json',
  },
  body: JSON.stringify({
  "model": "omnihuman-v2:flow",
  "prompt": "a beautiful landscape",
  "opts": {
    "image_url": "https://example.com/image_url.png",
    "audio_url": "https://example.com/audio_url.mp3"
  }
}),
});

if (!response.ok) throw new Error(`GetVibe API: ${response.status}`);
const { jobId } = await response.json();
console.log(jobId);