SuperAPI AI Directory

space-video-cover

Design & creation

View repository

Updated 10-16 07:23

About this skill

space-video-cover · Video Cover

The cover determines the click-through rate. The covers produced by this skill must satisfy two points:This one alone is eye-catching enough, andplaced in a row on the homepage, it's clear they're from the same account。

Family feel (core concept)

Don't lock into a rigid template, but hold to a set of "genes" — make your covers recognizable at a glance:

  • Fixed: primary/secondary colors, main font, title position area, account identifier position
  • Varies per piece: specific layout, imagery, accent color, information hierarchy

Redesign each time, maintain the family feel. Don't let ten covers look like they came from one mold, and don't let ten covers each say something different.

Platform ratios

Platform Ratio Suggested size
WeChat Channels / Douyin vertical 3:4 or 9:16 1080×1440 / 1080×1920
Bilibili horizontal 16:10 1280×800
Xiaohongshu 3:4 1080×1440
YouTube 16:9 1280×720
WeChat Official Account first image 2.35:1 1175×500

Ask the platform first, set the ratio, then design.

Design process

① 定信息层级:主标题(最大) + 副信息(次) + 视觉锚点(人/物/截图)
② 定钩子文案:封面标题 ≠ 视频标题,更短更冲,3-6 字爆点 + 补充
③ 排版:家族版式基因 + 本条强调色
④ 出图:审提示词 → 生成 → 检查可读性(缩略图大小下还能看清主标题吗)

Iron rules of readability

  • The main title must beat mobile list thumbnail sizeclearly readable at a glance (test by shrinking to 200px wide)
  • Enough contrast between text and background (light text on dark background/dark text on light background + stroke)
  • Don't cram in too much text, one cover one core point

Cover types

Type Suited for
Big-type poster (black-and-white/high-contrast text as the main element) Opinion type, suspense type
Person on camera + text Talking head, personal IP
Screenshot + highlighted annotations Hands-on tests, tutorials
Collage/editorial Roundups, comparisons

Orchestration (reuse existing capabilities)

  • Image generation and batch illustration → use the user's image-studio
  • Article/cover family and batch illustration → use the user's space-article-cover-and-batch-illustration
  • Xiaohongshu image-text cover → use the user's xiaohongshu-content-tools / related cover skills

This skill is responsible for defining the "family feel + information hierarchy + platform ratio + hook copy"; the actual image generation routes to the above skills.

Handoff

  • Upstream:space-video-topic the title/topic, or a frame screenshot from the finished video
  • Downstream:06_封面/cover.png → publish

Core principles

  • Family feel: recognizable > showing off on each piece
  • Thumbnail readable: the cover is for people who swipe past with their thumb in 0.5 seconds
  • Cover copy ≠ video title: shorter, punchier, hookier