make-explainer-video
Research, script, plan, generate, and assemble one short narrated explainer video with traceable facts, beat-timed visuals, subtitles, and intelligible audio. Use for educational mini-documentaries, t
SKILL.mdskills/make-explainer-video/SKILL.md
GitHubMake Explainer Video
Catalog
- Display name: Make Explainer Video
- Categories: Video
- Status: Ready
- Route label: Source-backed beat-to-video composition
- Tone: purple
- Short description: Produce a fact-traceable narrated explainer with timed visual beats, subtitles, and final QC.
- Similar skills: line-doodle-explainer, editorial-cutout-explainer
What this skill does
- Builds a source-backed explanation around one audience question and a clear narrative throughline.
- Locks narration timing before producing visually consistent beat assets.
- Assembles accepted clips with supplied audio without copyright authorization, subtitles, and a final sync/legibility review.
How to use
Provide the topic or source material, audience, angle/question, language, target duration, ratio, citation expectations, visual direction, and any supplied narration, music, brand, or reference assets.
Example
Make a 60-second 16:9 explainer about why urban trees cool streets. Use current primary sources, neutral paper-collage visuals, supplied narration voice, burned-in English subtitles, and source notes.
Workflow
- Lock topic, audience question, scope, duration, language, ratio, factual threshold, citation placement, tone, audio sources, and prohibited claims. Research current primary/authoritative sources when facts may have changed.
- Create a fact ledger linking every consequential number, date, quote, causal claim, and visual evidence claim to a source. Separate sourced fact, inference, and illustrative metaphor.
- Write an exact-duration beat script: concrete hook, necessary context, escalating explanation, complication, and resolution that reframes the opening. Default to one idea and one narration block per 6–12 second beat; remove throat-clearing.
- Obtain narration timing before visual generation. Use user-supplied narration or another supported audio route without copyright authorization; standalone voice/music generation is not implied by this Skill. Record measured duration per beat.
- Select an explanation-native treatment: diagram reveal for systems and sequence, whiteboard for hand-drawn teaching, isometric scenes for spatial/process relationships, data-led recap for personal metrics, or a custom documentary system. Define palette, framing, medium, typography intent, motifs, and one anchor frame. Do not imitate a named publisher, artist, or protected program package. Approve the anchor before remaining images.
- Generate one keyframe and, when needed, one restrained video clip per beat using the appropriate existing image/video Atom. Keep one stable operation key per asset, poll to terminal state, and inspect each result. Motion should explain the beat rather than decorate it.
- Conform only accepted clips to measured narration windows and assemble them in beat order. Add exact subtitles, citations, title cards, narration, supplied/approved music, and audio ducking through deterministic editing. Never ask a generative image/video model to guarantee long exact text.
- Watch the complete export and inspect representative beat frames as an ordered set. Confirm every diagram arrow/node, whiteboard stroke order, isometric relation, or data label communicates the scripted claim. Check source alignment, continuity, clip/voice/subtitle sync, safe zones, intelligibility, music level, duration/ratio, looping/drift, credits, and corrupt frames. Repair the originating stage only.
User-facing output
- Media type: One final MP4 plus fact ledger, beat script, source notes, and operation receipt
- Default quantity: 1
- Content per video: One researched explanatory arc with synchronized narration, visuals, and subtitles
- Default layout: User-requested ratio and exact duration target
- Model policy: Existing verified image/video routes; audio only when supplied or separately supported and supplied
- Downstream use: Human editorial, factual, and publication review