1. Text-to-Video

POST /text2video

Generate Videos Based on Text Prompts.

Request Body

{
  "prompt": "A Cat Running on the Grass",
  "size": "832*480",
  "base_seed": -1,
  "frame_num": 25,
  "sample_shift": 0.0,
  "sample_solver": "unipc",
  "sample_steps": 30,
  "sample_guide_scale": 5.0,
  "offload_model": false
}

Parameter Description

• prompt（Required）：Text Prompt for Video Generation

• size（Optional）：Dimensions of the generated video (widthheight), defaulting to "832*480" for image-to-video tasks

• base_seed（Optional）：Random seed for generating images or videos, with a default value of -1

• frame_num（Optional）：Number of frames sampled from images or videos, which should be a number in the form of 4n+1

• sample_shift（Optional）：Sampling offset factor of the flow matching scheduler

• sample_solver（Optional）：Sampling solver used, default is "unipc"

• sample_steps（Optional）：Number of sampling steps

• sample_guide_scale（Optional）：Classifier-free guidance scale, default value is 5.0

• offload_model（Optional）：Whether to offload the model to the CPU after each model forward propagation (to reduce GPU memory usage). The default value is false.

Response (200)

{
  "id": "Task ID",
  "status": "Task Status",
  "message": "Status Information"
}

2. Get Task Status

GET /status/{id}

Query the Status of a Specific Task.

Path Parameter

• id（Required）：Task ID

Response (200)

{
  "id": "Task ID",
  "status": "Task Status",
  "message": "Status Information"
}

3. Download Video

GET /download/{id}

Download the Generated Video.

Path Parameter

• id（Required）：Task ID

Response (200)

Video Data