| name | horizontal-diagnostic-talk-video |
| description | 横版诊断盘开场口播增强剪辑。先用全屏主题判断盘钉住问题，再切人物建立信任。
适合账号置顶视频、课程第一课、服务定位说明、业务场景诊断、方法论总览、"第一次来先看这条"。
当用户说"诊断盘开场"、"置顶视频"、"账号介绍视频"、"横版开场增强"、"主题先行"时使用。
|
Horizontal Diagnostic Talk Video
Goal
Create a 16:9 horizontal video that opens with a full-screen diagnostic panel, not the speaker. The viewer should understand the topic, audience, and judgment framework within 3 seconds — before the speaker ever appears.
This skill is for videos that answer "what should I look at first?" rather than "how do I do this?"
Required Inputs
- Source talking-head video (16:9)
- Original script, if available
- Diagnostic framework: categories, criteria, or judgment axes
- Title and CTA
- Output target:
1920x1080, 30fps, H.264/AAC
Workflow
- Inspect source with
ffprobe. Extract frames to check face position.
- Transcode to 1920x1080 H.264/AAC if needed.
- Transcribe and correct captions. Pay extra attention to judgment statements and CTA.
- Design the opening diagnostic panel: topic, audience, criteria, visual framework.
- Segment the rest into chapters: diagnostic opening → speaker explanation → flow boards → standards → CTA.
- Create props or edit timeline. The diagnostic panel should occupy 5-9 seconds at the start.
- Render stills for the diagnostic panel and key chapters.
- Render full video and generate contact sheet.
Layout Rules
Diagnostic Opening (5-9s)
- No speaker in first frame. Full-screen information panel.
- Topic title: large, centered, immediately readable.
- Audience labels: 2-4 tags showing who this is for.
- Framework axes: business categories, size ranges, readiness levels.
- Accent color: bold, distinguishing from later chapters.
- Transition into speaker: smooth, not abrupt.
Speaker Chapters
- Speaker becomes primary visual after diagnostic opening.
- Same rules as horizontal-content-talk-video: face safe, cards on sides, subtitles bottom.
- Use PIP when showing methodology boards.
- Return to full speaker for identity statements and CTA.
Flow Board Chapters
- Full-screen animated boards with PIP speaker anchor.
- Components reveal by phrase, not all at once.
- Board background: blurred source video for continuity.
Scene Pattern
diagnostic: Full-screen panel. No speaker. Topic + audience + framework.intro: Speaker enters. Restate the promise from the panel.mistakes: Full board + PIP. What gets wrong.method: Full board + PIP. Process or framework.standards: Full speaker. Identity or judgment summary.cta: Board or speaker. Clear action.
Animation Rules
- Diagnostic panel: elements fade in sequentially within the first 3 seconds.
- Panel-to-speaker transition: under 20 frames.
- Card reveal: 8-10 frames.
- Component lead:
0.12-0.18s.
- Caption lead:
0.05s.
- Split any static visual over 14 seconds.
Caption Rules
- Same as horizontal-content-talk-video.
- Additional check: diagnostic panel text must match the first spoken sentence.
- The spoken "这条视频给谁看" must align with the audience tags on screen.
Quality Gates
Fail if any of:
- The first 3 seconds do not reveal topic and audience
- Speaker appears before the diagnostic message is clear
- Face blocked at any point
- Captions out of sync
- Board content unreadable
- Template debug words visible
- CTA does not match voice
Validation Commands
ffprobe -v error -show_entries format=duration -show_streams -of json "output.mp4"
ffmpeg -y -i "output.mp4" -vf "fps=1/5,scale=480:-1,tile=5x3" -frames:v 1 "contact.jpg"
ffmpeg -y -ss 0 -i "output.mp4" -t 5 -vf "fps=2,scale=480:-1,tile=5x2" -frames:v 1 "opening-check.jpg"
Reference
Read references/diagnostic-sop.md for the full SOP with detailed diagnostic panel design, chapter mapping, Remotion layout specs, and delivery checklist.
