majiayu000 Veo Video Generation
Generate AI videos through AceDataCloud's Google Veo API.
Authentication
export ACEDATACLOUD_API_TOKEN="your-token-here"
Quick Start
curl -X POST https://api.acedata.cloud/veo/videos \
-H "Authorization: Bearer $ACEDATACLOUD_API_TOKEN" \
-H "Content-Type: application/json" \
-d '{"prompt": "a whale breaching in slow motion at golden hour", "model": "veo-3", "wait": true}'
Models
| Model | Duration | Audio | Best For |
|---|---|---|---|
veo-2 |
5–8s | No | Fast, cost-effective generation |
veo-3 |
8s | Yes (native) | Full audiovisual generation |
veo-3.1 |
8s | Yes (native) | Latest model, highest quality |
Workflows
1. Text-to-Video
POST /veo/videos
{
"prompt": "cinematic aerial shot of the Northern Lights over Iceland",
"model": "veo-3",
"aspect_ratio": "16:9",
"duration": 8
}
2. Image-to-Video
Animate a still image into video.
POST /veo/videos
{
"prompt": "the scene gently comes to life with wind and subtle motion",
"image_url": "https://example.com/landscape.jpg",
"model": "veo-2",
"aspect_ratio": "16:9"
}
3. Upscale to 1080p
Convert a generated video to full 1080p resolution.
POST /veo/videos/1080p
{
"video_url": "https://example.com/generated-video.mp4",
"model": "veo-3"
}
Parameters
| Parameter | Values | Description |
|---|---|---|
model |
"veo-2", "veo-3", "veo-3.1" |
Model to use |
aspect_ratio |
"16:9", "9:16" |
Video aspect ratio |
duration |
5 – 8 |
Duration in seconds |
generate_audio |
true / false |
Enable/disable audio (veo-3, veo-3.1 default to true) |
enhance_prompt |
true / false |
Let the model expand your prompt for better results |
Task Polling
POST /veo/tasks
{"task_id": "your-task-id"}
States: pending → succeeded or failed.
MCP Server
pip install mcp-veo
Or hosted: https://veo.mcp.acedata.cloud/mcp
Key tools: veo_generate_video, veo_generate_video_from_image, veo_get_1080p_video
Gotchas
- Veo 3 and 3.1 generate native audio — use
generate_audio: falseto suppress - The 1080p endpoint (
/veo/videos/1080p) only upscales previously generated videos enhance_promptcan significantly improve results but may deviate from literal interpretation- Veo 2 does NOT support audio generation
- Duration is capped at 8 seconds for all models
- Task states use
"succeeded"(not "completed") — check for this value when polling