Skip to content

Vega Video 1.0 Pro โ€‹

Updated

โ† Back to the Vega Video overview for version comparison and selection.

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

Vega Video 1.0 Pro (1.0-pro, plus 1.0-pro-fast / 1.0-pro-mini) is for video generation: text-to-video, first-frame / last-frame / first-and-last-frame generation, and multimodal reference generation from up to 9 images, 3 videos and 3 audio clips. It also supports editing and extending a reference video, outputs 480Pโ€“4K at 4โ€“15 seconds per clip, and can generate audio.

Basic Information โ€‹

ItemValue
ModelNamewand-vega-video
ModelVersion1.0-pro / 1.0-pro-fast / 1.0-pro-mini
Resolution480P / 720P / 1080P / 2K / 4K
Aspect ratio21:9 / 16:9 / 4:3 / 1:1 / 3:4 / 9:16 / adaptive
Duration4~15 seconds
Output soundSupported (AudioGeneration: Enabled)
Guardrail bypass supportSupported

Input Specifications โ€‹

1.0 Pro Reference Images โ€‹

ItemSpecification
Number of reference imagesUp to 9
Size per image fileLess than 30MB
Image width / height300~6000px each
Image aspect ratioWidth/height 0.4~2.5
Image formatsJPEG / PNG / WEBP / BMP / TIFF / GIF / HEIC / HEIF
First / last frame1 each
Total mixed inputsUp to 12 for images / videos / audio combined

1.0 Pro Reference Video / Audio โ€‹

ItemSpecification
Number of reference videosUp to 3
Size per video file50MB or less
Video duration2~15 seconds each, 15 seconds or less in total
Video width / height300~6000px each, width/height ratio 0.4~2.5
Video format / codecMP4 / MOV
Video resolution / total pixels480P / 720P / 1080P, 409,600~2,086,876 pixels
Video frame rate24~60fps
Number of reference audio filesUp to 3
Size per audio file15MB or less
Audio duration2~15 seconds each, 15 seconds or less in total
Audio formatsWAV / MP3
Audio input conditionMust be submitted together with 1 or more images or videos. Audio cannot be submitted on its own

Request Parameters โ€‹

ParameterRequiredTypeDescription
ModelNameRequiredStringFixed value wand-vega-video
ModelVersionRequiredString1.0-pro / 1.0-pro-fast / 1.0-pro-mini
PromptConditionalStringRequired when FileInfos is empty
FileInfos.NOptionalArrayReference inputs.
Type : String - input file type - File : VOD media file - Url : accessible URL
Category : String - file category - Image, Video, Audio
FileId : String - media file ID, valid when Type is File
Url : String - accessible file URL, valid when Type is Url
Usage is FirstFrame / LastFrame / Reference.
OutputConfig.ResolutionOptionalString480P / 720P / 1080P / 2K / 4K
OutputConfig.DurationOptionalFloatA number in seconds, range 4~15. -1 means the model determines the duration automatically.
OutputConfig.AspectRatioOptionalString21:9 / 16:9 / 4:3 / 1:1 / 3:4 / 9:16 / adaptive
OutputConfig.StorageModeOptionalStringTemporary / Permanent (storage type: temporary storage / permanent storage)
OutputConfig.MediaNameOptionalStringMedia name shown when Permanent
OutputConfig.AudioGenerationOptionalStringEnabled / Disabled
OutputConfig.LogoAddOptionalStringEnabled / Disabled (whether to apply the logo watermark)
SeedOptionalIntegerModel seed. The same Seed yields similar results
ExtInfoOptionalStringJSON encoding within 2. Passes additional options to AdditionalParameters / WandVegaParameters

Output Resolution โ€‹

Specified via OutputConfig.Resolution and AspectRatio. Output sizes based on 16:9 are as follows.

ResolutionAspectRatioOutput resolution
480P16:9854x480
720P16:91280x720
1080P16:91920x1080
2K16:92560x1440
4K16:93840x2160

480P is available on 1.0-pro.

Request Examples โ€‹

Text-to-Video โ€‹

json
{
  "SubAppId": 123456789,
  "ModelName": "wand-vega-video",
  "ModelVersion": "1.0-pro",
  "Prompt": "a calico cat sitting on a wooden windowsill, sunlight streaming in, gentle breeze",
  "OutputConfig": {
    "Resolution": "1080P",
    "Duration": 5,
    "AspectRatio": "16:9",
    "StorageMode": "Temporary"
  }
}

First Frame Generation (FirstFrame) โ€‹

json
{
  "SubAppId": 123456789,
  "ModelName": "wand-vega-video",
  "ModelVersion": "1.0-pro",
  "FileInfos": [
    {
      "Type": "Url",
      "Category": "Image",
      "Usage": "FirstFrame",
      "Url": "https://<cdn>/first.png"
    }
  ],
  "Prompt": "subtle camera dolly, calm mood",
  "OutputConfig": {
    "Resolution": "1080P",
    "Duration": 5,
    "AspectRatio": "16:9",
    "StorageMode": "Temporary"
  }
}

Last Frame Generation (LastFrame) โ€‹

Use Usage="LastFrame", or LastFrameUrl which is kept for compatibility with legacy calls.

json
{
  "SubAppId": 123456789,
  "ModelName": "wand-vega-video",
  "ModelVersion": "1.0-pro",
  "FileInfos": [
    {
      "Type": "Url",
      "Category": "Image",
      "Usage": "LastFrame",
      "Url": "https://<cdn>/last.png"
    }
  ],
  "Prompt": "the scene settles on a quiet window at dusk",
  "OutputConfig": {
    "Resolution": "1080P",
    "Duration": 5,
    "AspectRatio": "16:9",
    "StorageMode": "Temporary"
  }
}

First & Last Frame Generation (FirstFrame + LastFrame) โ€‹

json
{
  "SubAppId": 123456789,
  "ModelName": "wand-vega-video",
  "ModelVersion": "1.0-pro",
  "FileInfos": [
    {
      "Type": "Url",
      "Category": "Image",
      "Usage": "FirstFrame",
      "Url": "https://<cdn>/first.png"
    },
    {
      "Type": "Url",
      "Category": "Image",
      "Usage": "LastFrame",
      "Url": "https://<cdn>/last.png"
    }
  ],
  "Prompt": "smooth transition between the two scenes",
  "OutputConfig": {
    "Resolution": "1080P",
    "Duration": 5,
    "AspectRatio": "16:9",
    "StorageMode": "Temporary"
  }
}

Responses and Result Query โ€‹

Task Acceptance Response โ€‹

json
{
  "TaskId": "<task-id>",
  "RequestId": "<request-id>"
}

Result Query โ€‹

json
{
  "SubAppId": 123456789,
  "TaskId": "<task-id>"
}

Completion Response โ€‹

json
{
  "AigcVideoTask": {
    "Status": "FINISH",
    "ErrCode": 0,
    "Progress": 100,
    "Output": {
      "FileInfos": [
        {
          "FileUrl": "http://<host>.vod2.myqcloud.com/.../aigcVideoGenFile.mp4",
          "ExpireTime": "2026-09-15T10:29:48Z",
          "MetaData": {
            "Width": 1920,
            "Height": 1080,
            "Duration": 5.07,
            "Container": "mov,mp4,m4a",
            "Bitrate": 9494850
          }
        }
      ]
    }
  }
}

Special Settings โ€‹

ExtInfo (AdditionalParameters) โ€‹

Pro's ExtInfo uses either the AdditionalParameters or WandVegaParameters key depending on the option. It is JSON-encoded within 2, and the inner object is wrapped as a string once more.

  • return_last_frame : also returns the last frame image in the result.
  • bitrate_mode : high-quality mode
python
import json
ext_inner = {
    # "return_last_frame": True,   # also return the last frame image in the result
    # "bitrate_mode": "high",      # high quality mode
}
ext_info = json.dumps({"AdditionalParameters": json.dumps(ext_inner)}, ensure_ascii=False)

Example payload (retrieving the last frame):

json
{
  "SubAppId": 123456789,
  "ModelName": "wand-vega-video",
  "ModelVersion": "1.0-pro",
  "FileInfos": [
    {
      "Type": "Url",
      "Category": "Image",
      "Usage": "Reference",
      "Url": "https://<cdn>/ref.png"
    }
  ],
  "Prompt": "natural motion, cinematic",
  "OutputConfig": {
    "Resolution": "1080P",
    "Duration": 5,
    "AspectRatio": "16:9",
    "StorageMode": "Temporary"
  },
  "ExtInfo": "{\"AdditionalParameters\":\"{\\\"return_last_frame\\\":true}\"}"
}

When called with return_last_frame=true, Output.FileInfos contains both the video and the last frame image. The UsageType of the last frame is last_frame_url.

Task Type Selection โ€‹

Specify omni_reference_task_type in AdditionalParameters inside ExtInfo.

ValueBehaviorInput and output conditions
autoDetermines the task type automatically from the input material and the promptThe input and output specifications of each version apply
referenceGenerates a new video by referencing images / video / audioNo additional aspect ratio / duration constraints are imposed by task type. The base specifications of each version still apply
editEdits the picture or audio of the source video1 or more reference videos, video duration 4~30 seconds. OutputConfig.AspectRatio=adaptive, OutputConfig.Duration=-1
extendExtends the source video forward or backward1 or more reference videos. OutputConfig.AspectRatio=adaptive

Reference videos are passed via Category=Video and Usage=Reference of FileInfos. The video input duration and file limits of each version also apply.

python
import json

ext_info = json.dumps({
    "AdditionalParameters": json.dumps({
        "omni_reference_task_type": "extend"
    })
})

Model Options (WandVegaParameters) โ€‹

ParameterTypeValue / behavior
EnhancePromptModeStringauto / enable / disable. When omitted or set to auto, Pro treats it as disable. enable enables optimization, and disable uses the original prompt
ModerationModeStringModeration_L1: basic moderation, the default. Moderation_L2: enhanced moderation
ReturnVegaUsageBooleanDefault false. When true, usage is returned in the result

Pro and Lite differ in the default behavior of EnhancePromptMode. ModerationMode is an option for choosing the moderation level, not a guardrail bypass option.

python
import json

ext_info = json.dumps({
    "WandVegaParameters": json.dumps({
        "EnhancePromptMode": "enable",
        "ModerationMode": "Moderation_L2",
        "ReturnVegaUsage": True
    })
})

Pass the ext_info string in ExtInfo of the generation request. When using AdditionalParameters and WandVegaParameters together, turn each into a JSON string first, then put them into a single ExtInfo object and serialize it again.

In the usage response, Usage.CompletionTokens is the number of output tokens and Usage.TotalTokens is the total number of tokens including the input.

Seed โ€‹

Supplying the same Seed value produces similar results. Pin Seed when you need stable result comparison across regenerations.

json
{
  "SubAppId": 123456789,
  "ModelName": "wand-vega-video",
  "ModelVersion": "1.0-pro",
  "Prompt": "a calico cat on a wooden windowsill",
  "Seed": 12345,
  "OutputConfig": {
    "Resolution": "1080P",
    "Duration": 5,
    "AspectRatio": "16:9",
    "StorageMode": "Temporary"
  }
}

Live-Action Person Reference โ€‹

Live-action person reference can be integrated in two ways. Both are supported on Pro only, not on Lite:

  • Register as a material, then reference it (recommended): register the image with CreateAigcMaterial, get its AssetId, and reference it in FileInfos as asset://<AssetId>. Real-person materials on the China site require a liveness check first. See Real-Person Material Integration for the full procedure, and Non-Real-Person and International Site Materials for other materials.
  • Direct material group referencing: skip material registration and pass asset_group_id in the video generation request, as shown below.

Put the image URL into FileInfos of CreateAigcVideoTask, and pass asset_group_id to ExtInfo.AdditionalParameters. A group name used for the first time is generated automatically.

json
{
  "SubAppId": 123456789,
  "ModelName": "wand-vega-video",
  "ModelVersion": "1.0-pro",
  "Prompt": "natural motion, soft light",
  "FileInfos": [
    {
      "Type": "Url",
      "Category": "Image",
      "Usage": "Reference",
      "Url": "https://<cdn>/person.jpg"
    }
  ],
  "OutputConfig": {
    "Resolution": "720P",
    "Duration": 5,
    "AspectRatio": "16:9",
    "StorageMode": "Temporary"
  },
  "ExtInfo": "{\"AdditionalParameters\":\"{\\\"asset_group_id\\\":\\\"<your-group-id>\\\"}\"}"
}

Direct material group referencing is handled with a single CreateAigcVideoTask.

Permanent Storage โ€‹

json
{
  "OutputConfig": {
    "StorageMode": "Permanent",
    "MediaName": "wand-vega-video-pro-sample"
  }
}

Integration Settings โ€‹

ItemConfiguration basisHow to apply
Live-action person referenceRegister as a material and reference via asset://, or reference a material group directlyReal-person materials on the China site require a liveness check first. See Real-Person Material Integration
AudioAudio sources you own or have secured usage rights forSubmit it as reference audio, or generate the video only and combine them in the editing stage

Guardrail Bypass โ€‹

Input and output moderation can be configured separately.

Before using the guardrail bypass parameters, you must contact your sales representative to confirm your usage permissions. Even on supported engines, guardrails are bypassed only when the corresponding parameter is passed explicitly. Omitting the parameter leaves them in place.

ParameterBypass valueApplies to
OutputConfig.InputComplianceCheckDisabledBypasses input moderation
OutputConfig.OutputComplianceCheckDisabledBypasses output moderation

Merge the following settings into your existing generation request. To bypass both input and output moderation, pass both fields together.

json
{
  "OutputConfig": {
    "InputComplianceCheck": "Disabled",
    "OutputComplianceCheck": "Disabled"
  }
}

Built with VitePress ยท Deployed on Tencent EdgeOne Pages