viduq2-fast API 模型解析指南
快速结论
- Supports prompt-based text-to-image workflows for marketing and creative assets.
- Supports reference-guided generation for stronger style consistency.
- Stabilize parameters in this guide first, then move into production API.
核心能力
- High-quality generation：Generate production-ready visual assets via a unified API.
- Reference consistency：Use references to stabilize style and composition.
- Parameter tuning：Tune ratio/count/sampling for quality and cost balance.
适用场景
- When generating marketing, e-commerce, and social visual assets quickly.
- When reference-guided consistency is required.
不适用场景
- For long-form video generation workflows.
- For complex multi-turn reasoning conversations.
运行特性
- Uses unified image generation endpoints with task-style responses.
- Poll task status and fetch result URLs after completion.
最小请求示例
{
"model": "viduq2-fast",
"prompt": "Premium product poster, soft light, realistic texture, minimal background",
"size": "1:1"
}
最小响应示例
{
"id": "task_img_xxxxxxxx",
"object": "generation.task",
"model": "viduq2-fast",
"status": "queued",
"progress": 0
}
关键参数
| 参数 | 类型 | 必填 | 默认值 | 范围 | 说明 |
|---|
| model | string | 是 | viduq2-fast | - | Exact model identifier from ToAPIs. |
| prompt | string | 是 | - | - | Primary image instruction prompt. |
| size | string | 否 | 1:1 | - | Output aspect ratio, model-dependent. |
| n | integer | 否 | 1 | - | Number of generated images per request. |
| reference_images | file[] | 否 | - | - | Reference images for style consistency. |
常见错误
| HTTP | Code | 触发条件 | 修复建议 | 重试策略 |
|---|
| 400 | invalid_request_error | Missing or invalid payload fields. | Validate model/prompt/size payload fields. | Retry only after payload fix. |
| 401 | authentication_error | Missing or invalid API key. | Verify Authorization header and key scope. | Retry after authentication fix. |
FAQ
- What is viduq2-fast best for?
Best for high-frequency creative generation and reference-guided consistency workflows.
- 图像视频模型报错：invalid apitype: -1
这类错误通常说明接口走错了。图像和视频模型一般不走 chat 接口，而是按对应文档发起 HTTP 任务请求，并通过任务状态接口轮询结果。排查时建议先看用户的实际请求代码、请求地址和请求体。
- 用户进行生成图片/视频的任务时出现任务失败，但是扣款
先让用户提供任务日志或截图，重点看是否出现了输入或输出 token 统计。如果有这类 token 记录，大概率是用户把图片/视频模型走成了 chat 接口；这不是正确用法。图片和视频模型通常是异步任务接口，需要通过 HTTP 请求先提交任务，再拿到任务 ID 轮询状态，详细以对应文档为准。
相关 API
