Agent Skills
› 0xsline/OpenChatCut
› motion-graphic-placement
motion-graphic-placement
GitHub用于规划视频动态图形(MG)的插入时机、类型及布局,确保其增强叙事且不遮挡关键内容。通过分析脚本语义触发点,遵循放置规则优化视觉节奏与可读性。
Trigger Scenarios
用户希望为讲解、教程或访谈类视频添加动态图形以增强表现力
需要规划视频中动态图形的密度、类型及具体出现位置
Install
npx skills add 0xsline/OpenChatCut --skill motion-graphic-placement -g -y
SKILL.md
Frontmatter
{
"name": "motion-graphic-placement",
"description": "Add motion graphics at the right moments without blocking the story. Use when the user wants to enhance talking-head, lecture, tutorial, interview, podcast, or creator videos with motion graphics."
}
Motion Graphic Placement
Use this skill to plan where motion graphics should appear, what type each one should be, and how to keep them readable, rhythmic, and consistent with the video.
Goals
- Reinforce important speech moments without covering the speaker or subtitles.
- Match MG type to the semantic function of the line.
- Keep density high enough for retention but sparse enough to avoid fatigue.
- Maintain one visual language across the whole edit.
Workflow
- Read project state, transcript, active design style, video aspect ratio, and visible speaker layout.
- Identify the video structure:
- hook
- problem or setup
- body / proof / steps
- summary / CTA
- Decide motion-graphic density:
- long-form landscape: usually one meaningful MG every 5-8 seconds, tighter in the first 15 seconds
- short-form portrait: usually one meaningful MG every 3-5 seconds, with stronger rhythm
- reduce density when the speaker's face, gesture, or screen recording already carries the moment
- Pick MG types from semantic triggers.
- Review the plan before generation:
- no repeated form in a row unless intentionally templated
- no overlays on face, hands, product focal point, or subtitle area
- no unnecessary full-screen takeover when a card, strip, badge, or annotation is enough
- Generate or reuse MGs.
- Place each MG against the relevant transcript moment.
- Verify screenshots at target frames and fix anything that blocks content, feels mistimed, or carries template residue.
Semantic Triggers
| Speech pattern | Good MG direction |
|---|---|
| strong opinion, conclusion, key quote | highlight sentence or quote card |
| numbers, percentages, prices | number card, counter, chart, price card |
| steps, lists, sequence | point list, flow steps, timeline |
| comparison or choice | comparison card, split card, VS card |
| warning or mistake | warning card or caution badge |
| question, suspense, hook | question card, hook text |
| self-intro, guest, role | name tag or lower-third |
| source, screenshot, comment, evidence | source card or annotation |
| CTA | CTA card |
Placement Rules
- Always inspect the frame before placing an overlay.
- Place MGs away from the speaker's face, hands, active product area, and subtitles.
- Landscape: side cards work well when the speaker is clearly on the opposite side; bottom strips must sit above subtitles.
- Portrait: keep the bottom 15-20% clear for captions and interaction UI.
- Full-screen MGs should be reserved for hooks, chapter transitions, dense data, or moments where the original footage adds little value.
- After a full-screen MG, leave enough breathing room before the next overlay.
Style And Reuse
- Use the active Design Style when the project has one.
- Same-role MGs should reuse a confirmed template or code structure and change only the content.
- New-role MGs can borrow style, but should not inherit irrelevant template text, charts, labels, or layout.
- After generation, inspect the visual output. If old template text or wrong structure appears, fix the asset instead of trusting props.
Quality Checks
- Every MG has a reason tied to a speech beat.
- Text can be read at playback size.
- Adjacent MGs vary in form, position, or rhythm.
- No face/subtitle/product obstruction.
- Visual style is consistent but not repetitive.
- The final edit feels clearer, not busier.
Version History
- 0c0c02d Current 2026-08-12 09:07


