1
Enable checks
Request one to three supported safety categories on a single image or video job.
소스에서 결과물 전달까지
2
Layer the analysis
The pipeline performs a bounded first pass and escalates uncertain or higher-risk observations for deeper evaluation.
3
Apply your policy
Inspect the report and evidence in the webhook or result, then publish, hold for review, or reject in your application.
구축할 수 있는 기능
Image reports
Analyze one source image before its requested derivatives are produced and attach structured decisions to the job.
Video sampling
Sample frames at a service-controlled interval, aggregate the highest-risk observations, and return flagged-frame timestamps.
Structured evidence
Receive an overall verdict, flagged checks, per-check decisions and scores, plus bounded sampled-frame evidence.
Application-owned policy
Combine MediaRuntime’s report with human review, user history, appeals, and your product’s own publishing thresholds.
관련 프리셋
Works with image_multi_v1Works with video outputs such as mp4_720p_h264_aacChecks: sexual · violence · dangerousMode: report작업 요청 예제
신뢰할 수 있는 서버에서 MediaRuntime API 키로 제출하세요. 소스는 작업 시간 동안 접근 가능한 HTTPS URL이면 되며 MediaRuntime 업로드는 선택 사항입니다.
{
"file_url": "https://cdn.example.com/media/upload.mp4",
"metadata": {
"media_type": "video",
"asset_id": "asset-0426"
},
"moderation": {
"enabled": true,
"mode": "report",
"checks": [
"sexual",
"violence",
"dangerous"
]
},
"outputs": [
{
"type": "mp4",
"preset": "mp4_720p_h264_aac"
}
]
}