15+ Midjourney Prompts for YouTube Thumbnails That Get Clicks (2026)
Eighteen tested prompts for YouTube thumbnails — Midjourney reaction faces, versus splits and niche backgrounds, plus the same ideas rebuilt for Flux, SDXL and GPT Image 2 (formerly DALL·E 3). Each one is composed to stay readable at preview size, with an empty third saved for your title.
Viewers decide whether to click before they read a single word of your title. In the browse and suggested feeds where most views actually come from, the thumbnail is the entire pitch — rendered smaller than a postage stamp, competing against a dozen others on the same screen.
Our corpus holds 12 tested thumbnail prompts, every one at 16:9 — and across all 71 Midjourney entries, 72% carry a --no exclusion chain, with text and watermark the default exclusions for thumbnail work (full corpus data).
Key takeaways
- Always compose at 16:9 — all 12 of our tested thumbnail prompts use it, and cropping a square down destroys the composition.
- One focal subject, readable at 120px (roughly phone-feed preview size) — a second subject halves the impact of both.
- Reserve an empty third for your title and add the text in an editor — AI lettering comes out garbled, so
--no text, watermarkis the default exclusion. - Force high contrast with rim light and a complementary color pair (orange/teal, magenta/purple) so the face survives feed size.
This is exactly the kind of work AI image models are good at: dramatic backgrounds, exaggerated reaction bases, and concept shots (a glowing box, a rising arrow, a door you shouldn’t open) that would otherwise take an hour of stock-photo hunting or a paid designer. A round of Midjourney generations takes minutes.
Be honest about the limits, though. AI-rendered text comes out garbled more often than not, and a thumbnail with mangled lettering reads as spam. The workflow that works: generate the background and subject with AI, then add your title text, branding, and any real faces or product shots in an image editor. Every prompt below is written for that workflow.
The rules a thumbnail prompt must obey
Thumbnails have physics. Ignore these and even a beautiful render will underperform:
- 16:9, always. Every prompt here ends in
--ar 16:9(or states it for other models). Cropping a square render to widescreen destroys the composition. - One focal subject, readable at 120px. That’s roughly the preview size on a phone feed. One face, one object, one moment — a second subject halves the impact of both.
- An empty third for your title. Tell the model which part of the frame to leave clean — that’s where your editor text goes later.
- High contrast and rim light. A bright edge separating subject from background is what keeps a face legible at feed size. Complementary color pairs (orange/teal, magenta/purple) do the same job for the whole frame.
- Never ask the model to render text. Add
--no text, watermarkin Midjourney and composite your headline later. (The one exception is Flux — covered near the end.)
Browse every thumbnail-ready entry, with settings and example images, in our YouTube thumbnails use-case hub.
Reaction faces and creator portraits
Expressive faces are the highest-CTR thumbnail genre on YouTube, and Midjourney’s portrait quality makes them its home turf. The trick is prompting past polite stock-photo expressions into genuine exaggeration.
1. Shocked Scientist Thumbnail Background — Midjourney v7, --ar 16:9 --stylize 150
Our template portrait: an amazed expert on the left, title space on the right.
extreme close-up portrait of a scientist with a shocked amazed expression, safety goggles pushed up, glowing blue laboratory background, hard rim lighting separating subject from background, saturated complementary orange and teal, subject on the left third, empty space on the right for text, hyper-expressive youtube thumbnail style --ar 16:9 --stylize 150
Why it works: the composition instructions carry it — “subject on the left third, empty space on the right for text” reliably reserves your title area, and the hard rim lighting keeps the face readable at 200px. Swap “scientist” for any profession and the recipe holds.
2. Shocked Reaction Face Thumbnail — Midjourney v7, --ar 16:9 --stylize 200
When polite surprise isn’t enough, this one goes full jaw-drop.
close-up portrait of a man with an exaggerated shocked expression, jaw dropped, hands on cheeks, eyes wide, hot magenta rim light cutting around his silhouette, deep purple gradient background, high contrast studio lighting, subject filling the right two thirds, empty left third of the frame, punchy saturated color grade, hyper-expressive youtube reaction thumbnail style --no text, letters, watermark, logo --ar 16:9 --stylize 200
Why it works: “hands on cheeks, eyes wide” is the phrase that pushes the expression past stock-photo territory — drop it and you get a smile. If the subject drifts out of the reserved left third, reroll rather than fight it, and drop --stylize to 100 if skin starts looking airbrushed.
3. Podcast clip backdrop — Midjourney v7, --ar 16:9 --stylize 150
two microphones facing each other across a small studio table, warm tungsten key light, soft haze in the air, a single neon accent strip glowing in the dark background, microphones in the lower half of the frame, clean dark upper third, cinematic shallow depth of field --no text, watermark --ar 16:9 --stylize 150The clean upper third is the whole point — podcast clips live on a big quote overlay, and this frame hands you the space for it.
Versus energy for gaming channels
Gaming thumbnails reward conflict: two sides, two colors, one glance.
4. Gaming Versus Split Thumbnail — Midjourney v7, --ar 16:9 --stylize 250
The classic split-screen matchup, engineered so the halves stay distinct at feed size.
split-screen versus composition for a gaming thumbnail, left half an armored knight hero lit in fiery orange facing the center, right half a hooded rogue lit in electric blue facing the center, hard diagonal divide down the middle with a thin glowing seam, clean empty gap along the center line, dramatic game key art lighting, hyper-saturated contrasting halves, epic confrontation energy --no text, letters, numbers, logo, watermark --ar 16:9 --stylize 250
Why it works: lighting each half a different temperature — fiery orange against electric blue — does more for the versus effect than the divide itself, and the clean gap along the center line is where you drop your own VS badge in an editor. If one character bleeds across both halves, drop --stylize to 150 so Midjourney takes the divide literally.
5. Victory-scream gamer — Midjourney v7, --ar 16:9 --stylize 200
low-angle portrait of a gamer in glowing headphones roaring in celebration, fists raised, explosive burst of purple and cyan light behind his head, floating confetti particles, subject in the right two thirds, dark empty left third, hyper-saturated esports poster style --no text, watermark --ar 16:9 --stylize 200The light burst behind the head works as a built-in halo, separating the face from a busy background without needing a rim-light callout.
Money, tech and explainer niches
These niches punish clutter hardest — the winning thumbnails are almost minimalist, with one glossy object doing all the talking.
6. Floating phone tech review — Midjourney v7, --ar 16:9 --stylize 150
a sleek generic smartphone floating above an open upturned palm, cool cyan underglow, dark navy studio background, single dramatic spotlight from above, phone in the right third of the frame, clean empty left third, glossy commercial tech photography --no text, watermark --ar 16:9 --stylize 150Keep --stylize at 150 or lower so the phone stays an anonymous slab — higher values drift toward recognizable real-world designs you don’t want.
7. Coin jar with a green sprout — Midjourney v7, --ar 16:9 --stylize 150
a glass jar overflowing with plain golden coins on a clean desk, a single bright green sprout growing from the top of the pile, soft morning window light, mint and gold palette, jar in the right third, bright empty left third, crisp commercial still life --no text, numbers, watermark --ar 16:9 --stylize 150The sprout is the story — growth — and --no numbers stops Midjourney stamping garbled currency markings across the coin faces.
8. Glowing question mark — Midjourney v7, --ar 16:9 --stylize 120
a giant glowing hologram of a question mark floating above an open book on a dark wooden desk, warm desk lamp light from the side, deep teal background, floating dust particles, book and hologram in the right half, dark empty left half, clean 3d illustration style --no text, watermark --ar 16:9 --stylize 120A question mark is the one glyph Midjourney renders reliably — it’s a shape, not a word — which makes it the safest “curiosity” symbol for explainer channels.
Food, fitness and travel niches
Lifestyle niches all click for the same reason: a frozen split-second of motion.
9. Noodle pull close-up — Midjourney v7, --ar 16:9 --stylize 150
extreme close-up of chopsticks lifting glossy ramen noodles high above a steaming bowl, broth dripping mid-air, dramatic warm side lighting, rich amber and red color grade, dark blurred kitchen background, bowl in the lower right, dark empty upper left corner, editorial food photography --no text, watermark --ar 16:9 --stylize 150The mid-air noodle pull is the frozen-action moment that separates a thumbnail from a menu photo — keep some element suspended in every food prompt you write.
10. Kettlebell chalk slam — Midjourney v7, --ar 16:9 --stylize 180
a chalk-covered kettlebell slamming onto a dark gym floor, white chalk cloud exploding outward, single hard spotlight from above, deep blue shadows with one warm orange accent light, kettlebell in the lower right, dark empty upper left, gritty sports commercial photography --no text, watermark --ar 16:9 --stylize 180The chalk cloud gives a static object explosive motion, and the blue-shadow/orange-accent pair is the standard motivational color grade for a reason.
11. Cliff-edge traveler — Midjourney v7, --ar 16:9 --stylize 200
a traveler with a backpack seen from behind, standing at the edge of a cliff overlooking a turquoise bay at golden hour, warm rim light tracing the silhouette, saturated sunset palette, figure in the left third, open empty sky filling the right third, cinematic travel photography --no text, watermark --ar 16:9 --stylize 200Shooting the figure from behind sidesteps AI-face artifacts entirely and keeps the traveler anonymous enough to reuse across destination videos.
Curiosity niches: horror-lite and DIY
Two very different audiences, one shared mechanic — the viewer has to wonder what happens next.
12. The door at the end of the hallway — Midjourney v7, --ar 16:9 --stylize 150
a slightly open door at the end of a long dark hallway, cold green light spilling through the gap, faint mist along the floor, extreme contrast with deep shadows, door in the right third, near-black empty left third, suspenseful cinematic still, restrained pg-rated horror mood --no text, blood, gore, watermark --ar 16:9 --stylize 150The light through the gap does all the scaring — no monster required — which keeps the thumbnail advertiser-friendly while still raising neck hair.
13. Sawdust in the light — Midjourney v7, --ar 16:9 --stylize 150
close-up of hands in worn work gloves guiding a power drill into fresh pine wood, golden sawdust spraying through a beam of warm backlight, blurred pegboard of tools in the background, action in the lower right, soft empty upper left, sharp commercial craft photography --no text, watermark --ar 16:9 --stylize 150Backlit sawdust is the money detail here — it turns an ordinary workshop shot into something tactile enough to promise a satisfying build.
The same ideas on other models
The composition rules above — one subject, empty third, hard contrast, 16:9 — transfer to every model unchanged; only the dialect of the prompt shifts. If your channel workflow lives in another tool, these five entries rebuild the same thumbnail logic there.
14. Bold Text YouTube Thumbnail — Flux.1 dev, 28 steps, guidance 3.5
A high-energy YouTube thumbnail background: an exploding desktop computer with dramatic orange sparks flying against a deep teal gradient background, motion blur on debris, bold white 3D headline text reading "I BROKE IT" with a black outline on the left side, hyper-saturated colors, sharp studio lighting on the subject.
Why it works: Flux is the only model in our library we trust with in-image headline text — keep it to three short words in quotes, all caps, and describe the treatment (“bold white 3D with black outline”) or you get thin default lettering. Misspelled renders are seed luck; regenerate.
15. Glowing Tech Unboxing Thumbnail — Flux.1 dev, 28 steps, guidance 3.5
A photorealistic YouTube thumbnail background: two hands lifting the lid off a matte black gadget box, a burst of warm golden light escaping from inside the box and illuminating the fingers from below, faint dust particles floating in the beam, deep charcoal studio backdrop, the box positioned in the right half of the frame with a dark uncluttered left half, crisp detail on the hands and box edges, dramatic high-contrast commercial lighting.
Why it works: making the box itself the light source under-lights the hands in a way no external lamp can fake, and Flux follows the “dark uncluttered left half” spatial instruction reliably at cfg 3.5. If knuckles render warped, regenerate — the hands are half the story.
16. Sizzling Pan Cooking Thumbnail — SDXL 1.0, 30 steps, CFG 6.5
extreme close-up of a cast iron pan sizzling over a blue gas flame, diced red peppers, onion pieces and herb leaves frozen mid-air above the pan, oil droplets splashing, backlit steam rising, saturated warm color grade, dark blurred kitchen background, food composition in the lower right of the frame, dark empty upper left corner, high-speed food photography, razor-sharp detail
Why it works: “frozen mid-air” plus “high-speed food photography” is what stops SDXL smearing the flying ingredients into streaks, and if the food turns plastic, lower cfg toward 5.5 before touching steps — over-guidance is what glazes food.
17. Finance Growth Arrow Thumbnail — GPT Image 2 (ChatGPT), 16:9
A high-energy YouTube thumbnail background: a single glossy emerald-green 3D arrow rocketing diagonally upward from the lower left to the upper right, leaving a subtle motion trail, a pile of generic golden coins scattered at its base, clean soft-gradient mint background, dramatic studio lighting with a bright highlight on the arrowhead, the upper left of the frame kept clean and empty, no text or numbers anywhere in the image.
Why it works: GPT Image 2 desperately wants to write dollar signs on finance images, so “no text or numbers anywhere” is doing real work, and describing the coins as “generic golden” avoids garbled engravings on every face.
18. Fitness Transformation Gym Thumbnail — GPT Image 2 (ChatGPT), 16:9
A cinematic YouTube thumbnail background: the dark silhouette of an athlete standing in a hazy gym, backlit by a single bright cold-white floodlight, chalk dust drifting through the light beam, a barbell resting at their feet, deep blue shadow tones with one warm orange accent light from the side, the athlete placed in the right third of the frame with a dark empty left side, moody motivational atmosphere, only a silhouette with no facial features visible, no text in the image.
Why it works: the anonymous silhouette means the thumbnail never contradicts your actual footage, and “chalk dust drifting through the light beam” gives GPT Image 2 something physical to render — it sells the atmosphere better than the word “moody” ever does.
Make these prompts yours
Three moves turn this list into your channel’s thumbnail system. First, swap along the three axes that matter: niche (the scientist becomes a mechanic, the ramen becomes tacos), emotion (shock becomes delight, suspense becomes triumph) and palette (match your channel branding, but keep the pair complementary or contrast collapses at feed size). The composition scaffolding — subject placement, empty third, rim light — stays untouched.
Second, A/B test. Generate two variants that differ in exactly one thing — expression, palette, or subject side — and run them through YouTube’s built-in thumbnail testing. One deliberate variable per test is how you learn what your audience actually clicks.
Third, composite. The strongest thumbnails on big channels are hybrids: an AI-generated background with a real cut-out — your face, the actual product, a genuine screenshot — layered on top in an editor. The prompts above already reserve the space, and a real face over an AI scene keeps the thumbnail honest to the video behind it.
If you want to get better at writing these from scratch, our guide to writing Midjourney prompts breaks down the layered structure every prompt here is built on, and the Midjourney prompt hub has the full tested collection beyond thumbnails.
What makes a YouTube thumbnail prompt work?
A working thumbnail prompt asks for exactly one focal subject showing one clear emotion, reserves an empty third of a 16:9 frame for title text, and forces high contrast — rim lighting and a complementary color pair — so the image stays readable at 120-pixel preview size. And it never asks the model to render text: headlines get added in an editor, where they’ll actually be spelled correctly. Every prompt in this list is just those five constraints wearing a different niche.
More guides like this
Liked these prompts? Get the free pack
Subscribe and get the free pack of 25 tested prompts — then one email a week with the best new tested prompts and model news. No spam, unsubscribe anytime.