Skip to content

Veo โ€‹

Updated

Endpoint: POST https://vod.intl.tencentcloudapi.comAction: CreateAigcVideoTask

Veo is a video generation model that supports text-to-video and image-to-video. omni supports multimodal reference generation from images and video, video editing, and state-preserving editing, with natively synchronized audio.

Basic Information โ€‹

ItemValue
ModelNameGV
ModelVersion3.1 / 3.1-fast / omni
Default valueModelVersion=3.1 / Resolution=1080P
Guardrail disable supportNot supported

Specifications Supported by Version โ€‹

VersionResolutionAspect ratioLength
Common480P / 720P / 1080P / 2K / 4Kโ€”โ€”

Request Parameters โ€‹

ParameterRequiredTypeDescription
ModelNameRequiredStringFixed value GV
ModelVersionOptionalString3.1 / 3.1-fast / omni
PromptRequiredStringGeneration prompt
FileInfos.NOptionalArrayReference input. Usage is FirstFrame (first frame) or Reference (reference)
OutputConfig.ResolutionOptionalString480P / 720P / 1080P / 2K / 4K
OutputConfig.DurationOptionalIntegerVideo length (seconds)
OutputConfig.AspectRatioOptionalString16:9 / 9:16 / 1:1, etc.

Request Example โ€‹

json
{
  "SubAppId": 123456789,
  "ModelName": "GV",
  "ModelVersion": "3.1",
  "Prompt": "a calm sunset over the ocean, cinematic",
  "OutputConfig": {
    "Resolution": "1080P",
    "Duration": 5,
    "AspectRatio": "16:9",
    "StorageMode": "Temporary"
  },
  "SessionContext": "job-001"
}

Response Example โ€‹

json
{
  "AigcVideoTask": {
    "Status": "FINISH",
    "ErrCode": 0,
    "Progress": 100,
    "Output": {
      "FileInfos": [
        {
          "FileUrl": "http://<host>.vod2.myqcloud.com/.../aigcVideoGenFile.mp4",
          "ExpireTime": "2026-08-01T10:29:48Z",
          "MetaData": {
            "Width": 1920,
            "Height": 1080,
            "Duration": 5.07,
            "Container": "mov,mp4,m4a",
            "Bitrate": 9494850
          }
        }
      ]
    }
  }
}

Omni Input Specifications โ€‹

Omni references up to 10 images and up to 3 videos / 10 seconds each. The aspect ratio is 16:9 / 9:16, the length is 3โ€“10 seconds, and the frame rate is 24fps. The original resolution is 720P, and 1080P / 2K / 4K are upscaled outputs.

Special Settings โ€‹

omni Version Specifications โ€‹

omni is the version in the GV family that supports native audio generation and state-preserving editing.

ItemValue
Generation modeText to video, image reference, video reference, video editing, state-preserving editing
Reference imageUp to 10
Reference videoUp to 3 / 10 seconds each
Prompt lengthUp to 20,000 characters
AudioThe model generates audio together with the video.
Aspect ratio16:9 / 9:16
Resolution720P / 1080P / 2K / 4K
Video length3~10 seconds
Frame rate24 fps

Prompt Image Markers โ€‹

omni is a marker inside the prompt that specifies the role of an image. Image reference numbers start from 0.

MarkerRole
<FIRST_FRAME>Use this image as the starting frame of the video
<IMAGE_REF_N>Use the Nth image as a style or subject reference
text
<FIRST_FRAME> a woman is walking

in the style of <IMAGE_REF_0> a woman <IMAGE_REF_1> is walking

The second example uses the first image as a style reference and the second image as a subject reference.

State-Preserving Editing (PreviousTaskId) โ€‹

With PreviousTaskId, it inherits the context of the previous result and edits it repeatedly with prompts. It reflects only the specified changes and keeps the remaining elements.

Create the first task.

json
{
  "SubAppId": 123456789,
  "ModelName": "GV",
  "ModelVersion": "omni",
  "FileInfos": [
    {
      "Type": "Url",
      "Url": "https://<cos>/f0.jpg"
    }
  ],
  "Prompt": "Two man are studying",
  "EnhancePrompt": "Enabled",
  "OutputConfig": {
    "StorageMode": "Temporary"
  },
  "SessionContext": "job-001"
}

Pass the returned TaskId into ExtInfo as its PreviousTaskId to continue editing.

json
{
  "SubAppId": 123456789,
  "ModelName": "GV",
  "ModelVersion": "omni",
  "Prompt": "Make the background invisible",
  "EnhancePrompt": "Enabled",
  "ExtInfo": "{\"AdditionalParameters\": \"{\\\"PreviousTaskId\\\": \\\"<previous-task-id>\\\"}\"}",
  "OutputConfig": {
    "StorageMode": "Temporary"
  },
  "SessionContext": "job-001"
}

State-Preserving Editing Works Only for Tasks Within 24 hours

PreviousTaskId specifies a task created within 24 hours after generation.

Other Prompt Uses โ€‹

PurposeHow to write the prompt
Single sceneThe default is multi-shot. If you want a single scene, state it in the prompt
Removing elementsSpecify directly the elements you want removed
Audio instructionsWrite the sounds or dialogue you want together in the prompt
Time-specific eventsDescribe the actions at specific points in order
Text in videoYou can specify characters to place inside the video
Meta promptYou can instruct video quality or composition principles

Built with VitePress ยท Deployed on Tencent EdgeOne Pages