Skip to main content

​
Overview

WaveAPI supports multiple video generation models, all accessible through the unified async task endpoint /v1/tasks. Video generation typically takes longer, so use polling or webhook callbacks to retrieve results.

​
Request Endpoint

POST https://qingbo.dev/v1/tasks

​
Common Request Parameters

ParameterTypeRequiredDescription
modelstringYesModel name
actionstringNoDefaults to "generate"
promptstringYesVideo description text
durationintegerNoVideo duration (seconds)
aspect_ratiostringNoAspect ratio, e.g., "16:9"
first_framestringNoFirst frame image URL (image-to-video)
last_framestringNoLast frame image URL (requires first_frame)
video_urlstringNoReference video URL
video_urlsarrayNoMultiple reference video URLs
image_urlstringNoReference image URL
image_urlsarrayNoMultiple reference image URLs
generate_audiobooleanNoWhether to generate audio simultaneously
watermarkbooleanNoWhether to add a watermark
callback_urlstringNoWebhook callback URL upon completion
extraobjectNoModel-specific pass-through parameters

​
Response Format

{
  "task_id": "task_xxx",
  "status": "completed",
  "result": {
    "videos": [
      {
        "url": "https://...",
        "expires_at": 1720000000
      }
    ]
  }
}

​
Supported Models

VEO3

Google video generation model

Sora2

OpenAI video generation model

Kling

Kuaishou Kling series

More Models

Hailuo, Doubao, Vidu, Wan, and more

​
Generation Modes

Most video models support the following modes:
ModeParametersDescription
Text-to-videopromptGenerate video from text description
Image-to-videoprompt + first_frame or image_urlGenerate video from an image
First/Last framefirst_frame + last_frameGenerate a transition video between specified start and end frames
Not all models support all modes. Refer to each model’s documentation for specific support details.