Video Generation

Create Vidu video generation tasks through GregAPI, supporting text-to-video, image-to-video, first/last-frame, and reference-to-video.

1. Overview

Mode	Path	Description
Text-to-video	`/vidu/ent/v2/text2video`	Generate video from a text description
Image-to-video	`/vidu/ent/v2/img2video`	Generate video from a single image
First/last frame	`/vidu/ent/v2/start-end2video`	Generate a transition video from a first frame + last frame
Reference-to-video	`/vidu/ent/v2/reference2video`	Generate video from reference images + text description

export BASE_URL="https://api.gregapi.com"
export TOKEN="oh-xxxxxxxxxxxxxxxx"

GregAPI endpoint: POST $BASE_URL/vidu/ent/v2/{action}
Vidu official endpoint: POST https://api.vidu.cn/ent/v2/{action}

All requests use the GregAPI API Token:

Authorization: Bearer <TOKEN>
Content-Type: application/json

2. Available models

Scenario	Models
Image-to-video	`viduq3-turbo`, `viduq3-pro`, `viduq2-pro-fast`, `viduq2-pro`, `viduq2-turbo`, `viduq1`, `viduq1-classic`, `vidu2.0`, `vidu1.5`
First/last frame	`viduq3-turbo`, `viduq3-pro`, `viduq2-pro-fast`, `viduq2-pro`, `viduq2-turbo`, `viduq1`, `viduq1-classic`, `vidu2.0`
Text-to-video	`viduq3-turbo`, `viduq3-pro`, `viduq2`, `viduq1`, `vidu2.0`, `vidu1.5`
Reference-to-video	`viduq2`, `viduq1`, `vidu2.0`, `vidu1.5`

3. Common parameters

Parameter	Type	Required	Description
`model`	string	✅	Model name
`duration`	int	❌	Video duration in seconds; varies by model
`resolution`	string	❌	Resolution: `540p` / `720p` / `1080p`
`movement_amplitude`	string	❌	Camera motion: `auto` / `small` / `medium` / `large`
`off_peak`	bool	❌	Off-peak mode; half price when enabled
`watermark`	bool	❌	Whether to add a watermark
`wm_position`	string	❌	Watermark position
`wm_url`	string	❌	Watermark image URL
`callback_url`	string	❌	Callback URL

4. Text-to-video

curl -X POST "$BASE_URL/vidu/ent/v2/text2video" \
  -H "Authorization: Bearer $TOKEN" \
  -H "Content-Type: application/json" \
  -d '{
    "model": "viduq3-turbo",
    "prompt": "A girl runs toward the camera and winks with a smile.",
    "audio": true,
    "duration": 5,
    "resolution": "720p",
    "style": "general",
    "off_peak": false
  }'

Mode-specific parameters:

prompt (required): prompt, up to 2000 characters
style: general (general) or anime (anime)
audio: whether video and audio are generated together (q3 series only)

5. Image-to-video

curl -X POST "$BASE_URL/vidu/ent/v2/img2video" \
  -H "Authorization: Bearer $TOKEN" \
  -H "Content-Type: application/json" \
  -d '{
    "model": "viduq3-turbo",
    "images": ["https://your.cdn.com/first.png"],
    "prompt": "The astronaut waved and the camera moved up.",
    "audio": true,
    "duration": 5,
    "resolution": "720p",
    "off_peak": true
  }'

Mode-specific parameters:

images (required): image URL array, must contain exactly 1 image

6. First/last-frame generation

curl -X POST "$BASE_URL/vidu/ent/v2/start-end2video" \
  -H "Authorization: Bearer $TOKEN" \
  -H "Content-Type: application/json" \
  -d '{
    "model": "viduq2-pro",
    "images": [
      "https://your.cdn.com/start.png",
      "https://your.cdn.com/end.png"
    ],
    "duration": 5,
    "resolution": "1080p"
  }'

Mode-specific parameters:

images (required): image URL array, must contain exactly 2 images (first frame + last frame)

7. Reference-to-video

curl -X POST "$BASE_URL/vidu/ent/v2/reference2video" \
  -H "Authorization: Bearer $TOKEN" \
  -H "Content-Type: application/json" \
  -d '{
    "model": "viduq1",
    "images": [
      "https://your.cdn.com/ref-1.jpg",
      "https://your.cdn.com/ref-2.jpg"
    ],
    "prompt": "A mouse runs toward the camera, smiling and blinking.",
    "duration": 5,
    "resolution": "1080p"
  }'

Mode-specific parameters:

images (required): reference image URL array, 1–7 images
prompt (required): prompt, up to 2000 characters

8. Response example

{
  "task_id": "176843862716480",
  "platform_id": "video_01JSGXXXXXXXXXXXXXXXXXX",
  "state": "created"
}

task_id / id is the Vidu upstream task ID and can be used directly with Vidu’s official query interface. platform_id is GregAPI’s unified tracking ID (video_<ULID>) and can be used for platform-side query and cancel operations. Query and cancel both accept either ID. A successful creation only means the task is enqueued; get the final result through the Task Query endpoint.

9. Notes

Off-peak mode: off_peak=true halves the price but may delay generation
Resolution and duration: the q3 series supports 1–16 seconds; the Q2 series supports 1–8 seconds; vidu2.0 supports 4s/8s; vidu1.5 does not support first/last frame
Async tasks: all generations are asynchronous; poll or set a callback to get results

Large Language Models (LLM)

Image Models

Video Models

Speech Models

Video Generation

1. Overview

2. Available models

3. Common parameters

4. Text-to-video

5. Image-to-video

6. First/last-frame generation

7. Reference-to-video

8. Response example

9. Notes

​1. Overview

​2. Available models

​3. Common parameters

​4. Text-to-video

​5. Image-to-video

​6. First/last-frame generation

​7. Reference-to-video

​8. Response example

​9. Notes

1. Overview

2. Available models

3. Common parameters

4. Text-to-video

5. Image-to-video

6. First/last-frame generation

7. Reference-to-video

8. Response example

9. Notes