YouTube thumbnail concept prompt library with Raphael AI https://raphael.app
Find a file
Repository files (latest commit first)
Filename Latest commit message Latest commit date
2026-10-23 15:52:45 +00:00
README.md Update README.md 2026-10-23 15:52:45 +00:00

YouTube Thumbnail Concept Prompts

Thumbnails decide whether a video gets clicked. Creators who test several concepts per upload often find large differences in click-through rate. This repo is a prompt library for drafting thumbnail backgrounds and visual hooks with Raphael AI, a free AI image generator at https://raphael.app.

Principles

  • One focal point, readable on a phone screen.
  • High contrast between subject and background.
  • Leave space for three to five words of text added later.
  • Use the creator's real face photo for faces; generate only backgrounds and props.

Background prompts by niche

  • Tech review: "sleek desk setup with neon rim light, dark gradient background, shallow depth of field"
  • Cooking: "rustic kitchen counter with flour dust and morning light, blurred background"
  • Travel: "dramatic mountain road at sunrise, vivid sky, wide angle"
  • Finance: "rising chart made of glowing blocks, deep green and gold palette"

Testing workflow

  1. Make three distinct concepts per video.
  2. Composite your photo and title in an editor.
  3. Use the platform's thumbnail test feature where available.
  4. Record which colours and layouts win so future uploads start from data.

Honesty

Thumbnails should reflect the video. Misleading imagery hurts retention and audience trust, and platforms may penalise it.