Skip to content

Vega Agent โ€‹

Updated

Endpoint: https://smartmedia.vod-qcloud.com/aguiAuthentication: VOD ApiToken / Bearer

It edits video assets in the selected order and conditions, and performs subtitle and voice processing plus rendering. If this is your first call, run the two-video concatenation example first.

Basic Information โ€‹

ItemValue
RequestPOST /agui/chat
Scenariovideo-mixcut
InputProduction instructions + video URL / VOD fileId
ResultEdited video / text and tool execution events

Supported Specifications by Version โ€‹

VersionModel ID
1.1 Maxwand-vega-agent-1.1-max
1.0 Maxwand-vega-agent-1.0-max
1.0 Prowand-vega-agent-1.0-pro
1.5 Standardwand-vega-agent-1.5-standard
1.0 Standardwand-vega-agent-1.0-standard
1.0 Litewand-vega-agent-1.0-lite

Production Modes โ€‹

ModeOriginal AudioScript / VoiceUse Case
Original AudioKeep all / Remove BGM and keep vocalsUse the original audioInterviews, reviews, live highlights
NarrationRemoveUser script / AI script + TTSProduct descriptions, explainers, introduction videos

Request Parameters โ€‹

ParameterRequiredTypeDescription
messages.contentRequiredArrayProduction instructions and asset list
content.typeRequiredStringtext for instructions, mixcut_assets for assets
content.textConditionally requiredStringEditing conditions such as order, length, and subtitles
content.metadata.attachmentsConditionally requiredArrayurl or fileId for each asset
forwardedProps.modelOptionalStringModel ID in the table above
forwardedProps.scenario_nameRequiredStringvideo-mixcut

Authentication, run ID, and approval configuration follow API Reference.

Request Example โ€‹

This is the request used in input videos A and B and the final video.

json
{
  "threadId": "thread-001",
  "runId": "run-001",
  "messages": [
    {
      "role": "user",
      "content": [
        {
          "type": "text",
          "text": "Concatenate the attached videos A and B in order to produce a 6-second landscape video. Place the full 3 seconds of B after the full 3 seconds of A and connect them directly with no transition effect. Keep the original audio of each video and do not add separate music or narration. Add \"Sweet World\" to segment A and \"Today's Star\" to segment B as centered subtitles at the bottom of the screen. Set the subtitles to Noto Sans KR, white 96px, black outline 4px. Render at 1920x1080 and return the finished video URL."
        },
        {
          "type": "mixcut_assets",
          "metadata": {
            "attachments": [
              {
                "url": ""
              },
              {
                "url": ""
              }
            ]
          }
        }
      ]
    }
  ],
  "forwardedProps": {
    "model": "wand-vega-agent-1.0-pro",
    "scenario_name": "video-mixcut",
    "approval_mode": "never"
  }
}

Save it as request.json and then run it.

bash
curl -N -sS https://smartmedia.vod-qcloud.com/agui/chat \
  -H "Authorization: Bearer $VOD_API_TOKEN" \
  -H "Content-Type: application/json" \
  --data-binary @request.json

Attaching a VOD File โ€‹

Instead of a URL, use the VOD file ID registered in the same application.

json
{
  "type": "mixcut_assets",
  "metadata": {
    "attachments": [
      {"fileId": "387702300001234567"},
      {"fileId": "387702300001234568"}
    ]
  }
}

Special Settings โ€‹

Write the following conditions in natural language in text.

ItemWhat to Specify
Purpose / AudiencePurpose of the video, target viewers, key message
StructureIntro, body, outro, and scenes that must be included
LengthAuto / 15 seconds / 20 seconds / 25 seconds / 30 seconds / 60 seconds / Custom
ScreenLandscape / portrait, converting landscape assets to portrait
SubtitlesNot inserted / inserted, font size and outline
Background musicNone / uploaded music / generated music
Original audioKeep all / Remove BGM and keep vocals / Remove
NarrationUser script / generated script, TTS voice

Keeping the Original Audio โ€‹

text
Select the segments from the interview footage that describe the product usage experience and edit them into a 30-second vertical video.
Preserve the speaking order and meaning, and remove unnecessary pause segments.
Keep the original audio and do not add background music.
Add Korean subtitles at the bottom of the screen.

Narration-Focused โ€‹

text
Produce a 30-second landscape video from the product introduction footage.
The structure is product introduction, two key features, then usage scenes.
Write a Korean script from the information confirmed in the footage and synthesize narration.
Remove the original audio and arrange the scenes to match the narration.
Add subtitles and do not add background music.

To run after reviewing the edit plan, set approval_mode to level:high. Approval and Resume

Built with VitePress ยท Deployed on Tencent EdgeOne Pages