Create Killer Thumbnails With This Midjourney Technique

  • Beginner
  • 4 steps
  • By The Rundown University

Learn a quick prompting and editing system for making thumbnail images that stand out without looking like AI slop. You will turn a content title into a Midjourney-generated thumbnail background, then finish it with clean text in Canva, Photoshop, or Figma. The source workflow is designed to get you from title to finished thumbnail in about two minutes.

Step by step 4 steps

  1. Step 1 of 4

    Get AI to generate prompt concepts

    Open Claude, ChatGPT, Gemini, or another AI chat assistant and start a new chat. Paste this prompt, replacing the title placeholder with your content title.

    Use the following formula and article title to create four prompt concepts I can put into Midjourney to generate a thumbnail I can add text to later. Make sure there's negative space for the text.
    
    Formula: [Person + Expression] + [Action/Prop] + [Setting] + [Lighting] + [Composition with negative space] --ar 16:9
    
    Title: [paste your title here]

    You should get four scene concepts based on your title. Each concept should create a real, physical moment rather than an abstract AI-looking background.

    • If you do not want a person in the thumbnail, remove “Person + Expression” from the formula and ask for concepts without people.
    • You can also ask for concepts without the formula. This can produce more creative ideas, and sometimes those unstructured ideas work better.
    • If the concepts are not focusing on the right thing, tell the assistant what the main visual hook should be. For example: “The key concept is translation between languages.”

  2. Step 2 of 4

    Generate the thumbnail background in Midjourney

    Go to the Midjourney web app and paste your first prompt into the imagine bar.

    • Load all four prompts from Claude so you have options. Some will miss, and some may surprise you.
    • When you see a generation you like, click it and try Vary Subtle for small tweaks or Vary Strong for bigger changes.
    • Keep going until you land on an image that feels right, then download it to your desktop.

    Do not force a weak concept. If none of the four concepts are clicking, go back to Claude and ask for four more.

    Pro tip: Click the Draft Mode lightning bolt in the top right of Midjourney to save tokens on generations.

  3. Step 3 of 4

    Add simple, readable text in Canva

    Open Canva and search for a YouTube Thumbnail template, or create a blank design at 1280×720. Drag your Midjourney image into the design and make it full size.

    Keep the text simple. Add 1 to 5 words of big, attention-grabbing text. This text should be different from your actual title; use an action, an outcome, or a reaction instead.

    For a video titled “Translate Any Video File Using This Free Local AI Model,” the thumbnail text might be “MAKE GLOBAL CONTENT,” “ANY LANGUAGE,” or “FREE TOOL.”

    image-2097
    • Pick a big, blocky font.
    • Use white or yellow text.
    • Add a black outline for contrast.
    • Bring the opacity down about 5% to blend it with the background.
    • If readability is still an issue, put a darker color box behind the text and turn the box transparency way down.
    Pro tip: If the text does not fit because there is not enough empty space, go back to Claude and ask for a new Midjourney prompt with more negative space where you want the text. This is easier than trying to force the layout afterward.

  4. Step 4 of 4

    Add depth with the background remover trick

    Use this layering move to make the thumbnail look more designed and add depth.

    • Take your Midjourney background image in Canva and duplicate it.
    • Put the duplicate directly on top of the original.
    • Use Background Removal on the top layer. This is a Canva Pro feature.
    • Arrange the design so you have three layers: the original image on the bottom, text in the middle, and the foreground cutout on top.

    This makes the text look like it is sitting between the foreground and background, which adds depth and makes the thumbnail feel more intentionally designed.

    • This works especially well when the image has a person in it. Put the text behind their head.
    • Do not use this on every image, and do not cover too much of the lettering.
    • The goal is to make the viewer’s brain work a little harder when it sees the thumbnail. That curiosity helps earn the click.

That's the whole workflow

Mark it complete to keep your progress. The background follows.

Keep learning with Rundown Pro

Guides stay free. This is what Rundown Pro adds:

  • Every course and recording
  • The Rundown Certified Operator program
  • Member perks

$29/mo, cancel anytime

Background and context

Who This Is Useful For

  • YouTube creators and podcasters who need a new thumbnail every week without hiring a designer.
  • Bloggers and newsletter writers who need hero images and are tired of stock photos.
  • Marketing teams and social managers producing thumbnails at scale without sacrificing quality.

What You Will Build

image-2096

You will build a repeatable thumbnail production system: generate thumbnail backgrounds in Midjourney, then edit them with strong, readable text in Canva, Photoshop, or Figma.

  • Midjourney thumbnail background prompts based on your content title.
  • A finished 16:9 thumbnail image with negative space for text.
  • A reusable text-and-depth editing approach for future thumbnails.

What You Need

  • A Midjourney account. The source notes that the $10/month starter plan works.
  • Claude or another AI chat assistant, such as ChatGPT or Gemini, for generating prompt concepts.
  • Canva, Photoshop, or Figma. Canva Free works, while Canva Pro is needed for the background remover trick.
  • A content piece that needs a thumbnail, such as a YouTube video, blog post, or social post.

Going Further

  • Build a style library in Midjourney. Drag a thumbnail you like into the Style Reference slot and pin it with the lock icon so new generations match that visual style. This is useful for keeping a YouTube series or blog visually consistent.
image-2098