This is the one-page reference for prompting Meta Muse Image — Meta Superintelligence Labs' agentic image model, released in July 2026 across the Meta AI app, meta.ai, Instagram Stories (US), and WhatsApp. Its paired reasoning layer, Muse Spark, plans the layout, pulls in real-time web context, blends reference photos, and self-revises before it draws. The key thing to know: Muse Image takes full natural-language sentences — no comma tags, no --flags. Below is the prompt formula and copy-paste tables for every modifier that reliably moves the output.

New to it? Start with the best Meta Muse Image prompts roundup, then use this sheet while you build your own. For a deeper walk-through, see how to prompt Meta Muse Image for photorealism.

Advertisement

The prompt formula

A strong Muse Image prompt is one or two descriptive sentences that walk through seven parts in plain prose. Cover subject and setting first; the rest sharpen the result. Because Muse Spark reasons over the whole sentence, the order is a guide, not a rule.

The universal formula, in order:

  1. Subject — who or what, described concretely (age, clothing, material, expression).
  2. Setting — where it happens, with real detail (time of day, weather, surfaces).
  3. Lighting — the single biggest lever on mood; name a real setup.
  4. Camera / lens — how the scene should frame, compress, and blur.
  5. Style — photoreal, cinematic, a named preset, or an art medium.
  6. Text — any words in the image, in "quotes," with a named font.
  7. Aspect ratio — stated in words, e.g. "a wide 16:9 landscape."

Skeleton to copy — write it as flowing prose, not a tag list:

A photo of [subject, described concretely] in [setting with real detail],
lit by [lighting setup], shot on [lens] with [depth of field], in a
[style or preset] look. Include the text "[exact words]" in a [named font].
Make it a [aspect ratio] frame.

The same skeleton, filled in:

A photo of a weathered fisherman in his sixties wearing a knitted grey
sweater, standing on a misty harbour wall at dawn, lit by soft golden-hour
backlighting with a gentle rim light, shot on an 85mm lens with a shallow
depth of field, in a photorealistic editorial look with natural skin
texture. Make it a 4:5 portrait frame.

Why it works: every part is present and specific, it reads as a sentence a photographer could shoot, and Muse Spark can plan a coherent layout from it — exactly what the model was trained to reward.

Prompt building blocks

Each row is one slot in the formula. Say the thing on the left in the plain-language phrasing on the right.

ElementWhat to sayExample phrase
SubjectName it concretely — age, clothing, material, expression"a young barista with a nose ring and a green apron"
SettingPlace plus time of day, weather, and surfaces"in a sunlit corner cafe on a rainy morning"
LightingName a real lighting setup"lit by soft window light from the left"
Camera / lensLens and depth of field, in words"shot on a 35mm lens with a shallow depth of field"
StyleOne clear look or a named preset"in a warm editorial-film look"
TextExact words in quotes, font named"the headline \"OPEN\" in a bold condensed sans-serif"
Aspect ratioState the frame shape in words"make it a 9:16 vertical frame"

Aspect ratios

Set the ratio from the control in the Meta AI app or on meta.ai, or just state it in words — there are no flags. Muse Image supports six ratios. Spelling it out in prose works everywhere, including WhatsApp, where there is no toggle.

RatioBest forHow to ask
1:1Profile pics, album art, Instagram grid, product tiles"make it a 1:1 square"
4:3Presentation slides, classic monitors, blog headers"use a 4:3 landscape frame"
3:2Classic photography, editorial, prints"a 3:2 photographic landscape"
16:9YouTube thumbnails, hero banners, desktop wallpaper"a wide 16:9 landscape"
9:16Stories, Reels, TikTok, phone wallpapers"a tall 9:16 vertical frame"
4:5Instagram feed portrait — the max vertical the feed allows"a 4:5 portrait frame"

Lighting modifiers

Lighting sets the mood faster than any other part. Name a real setup and Muse Image renders it convincingly. Drop one phrase into the lighting slot of the formula.

TermLook
Soft window lightGentle, directional daylight — flattering for portraits and food
Golden-hour backlightWarm glow and long shadows, a halo around edges — dreamy, cinematic
Three-point studio lightClean, even, flattering — the safe default for portraits and products
Overcast diffusionFlat, shadowless, gentle — natural skin, honest product colour
Rim lightBright edge that separates the subject from a dark background
Rembrandt lightingSmall triangle of light on the shadowed cheek — classic portrait look
High-key lightBright, airy, low-shadow — beauty, e-commerce, clean and optimistic
Low-key lightMostly dark with selective highlights — noir, luxury, tension
Neon practical lightsColoured urban glow (magenta, cyan) reflecting on wet surfaces
Hard midday sunCrisp, high-contrast shadows — bold fashion and street work
Candlelight / firelightWarm, flickering, low and intimate — cosy and romantic scenes

Camera & lens language

Real lens language tells Muse Spark how the scene should compress, blur, and frame. Say it in plain words inside the sentence ("shot on an 85mm lens").

TermEffect
85mm lensFlattering portrait compression with creamy background blur
35mm lensNatural, documentary, street perspective close to how the eye sees
24mm wide-angleExpansive scenes, interiors, landscapes — some edge stretch
Macro close-upExtreme detail — jewellery, food, textures, small objects
Shallow depth of fieldSharp subject, soft background — isolates and draws the eye
Deep depth of fieldEverything sharp front to back — landscapes, architecture
Soft bokehRound out-of-focus highlights behind the subject
Tilt-shiftSelective focus band that makes real scenes look miniature
Aerial / drone viewTop-down or high overhead view for scale and pattern
Low angleLooking up — makes the subject tower and feel heroic
Eye-levelNeutral, relatable framing — the everyday default
Advertisement

Style & preset keywords

The style part decides whether you get a photo, a render, or an illustration. Tap a built-in preset or name the look in words; mixing three fights the model. The presets below ship inside Muse Image.

Preset / styleResult
Editorial filmMagazine-grade colour grade, natural texture, moody contrast
Renaissance portraitOld-master painting — soft chiaroscuro, rich fabric, classical pose
Retro posterMid-century print look — bold shapes, limited palette, grain
ClaymationStop-motion clay figures with fingerprint texture and soft studio light
16-bit video gamePixel-art sprites and tiles with a retro console palette
Restore old photoRepairs a faded or torn upload — recovers detail and colour
Trending hairstyleSwaps in a current cut or colour on a person you upload
PhotorealisticLooks like a real photograph — natural texture and lighting physics
CinematicFilm colour grade, wide-frame drama, atmospheric haze
Flat vector illustrationBold shapes, limited palette, no gradients — icons and web art
WatercolourSoft washes, bleeding edges, visible paper texture

Editing & reference actions

Muse Image is agentic, so editing and blending are first-class. If an image comes back about 80 percent right, don't regenerate — circle the one area to fix or reply with the single change, and Muse Spark keeps the rest. For ready-made skeletons, see the Meta Muse Image prompt templates.

ActionHow to phrase it
Region editCircle or sketch the area, then: "redraw only this — make the jacket red leather"
Conversational refine"keep everything else the same; make the sky overcast"
Blend references"@ mention photo 1 and photo 2 and combine them into one scene"
Character consistency"use photo 1 for the person's face across all three shots"
Product consistency"keep the exact bottle from photo 1, change only the background"
Style transfer"apply the painting style of photo 2 to the portrait in photo 1"
Add legible text"add the headline \"SALE\" in a bold condensed sans-serif at the top"
Working QR code"add a scannable QR code linking to https://memons.ai in the corner"

Muse Image vs flags: there are no --parameters in Muse Image — no --ar, --style, --v, and no numeric flags of any kind. Everything is plain prose. Write "a wide 16:9 landscape at golden hour," not "--ar 16:9". Comma-separated tag-soup and Midjourney-style switches actually hurt results, because Muse Spark is reading and reasoning over your sentence, not parsing tokens.

5 example prompts

Five complete prompts that assemble the modifiers above. Paste any into the Meta AI app or meta.ai and tweak the bracketed parts. For more, browse the best Meta Muse Image prompts.

1. Product hero shot

A photo of a matte-black ceramic coffee mug with steam rising, centred on a
wet slate surface in a dim minimalist studio, lit by a three-point studio
light with a subtle rim light, shot as a macro close-up with a shallow depth
of field, in a photorealistic product look with crisp reflections and a
seamless charcoal backdrop. Make it a 1:1 square.

Best for: e-commerce listings and ads — the studio light and macro framing make the material read as premium.

2. Cinematic environment

A photo of a lone figure in a long coat standing mid-stride at a rain-slicked
neon crossroads in a futuristic Tokyo alley at night, lit by magenta and cyan
neon practical lights reflecting off the wet asphalt, shot on a 35mm lens with
a deep depth of field from a low angle, in a cinematic look with atmospheric
haze. Make it a wide 16:9 landscape.

Best for: moody wallpapers and story frames — the neon lighting and low angle do the cinematic heavy lifting.

3. Editorial portrait

A photo of a confident woman in her thirties with short curls and a tailored
linen blazer, looking straight at camera in a sunlit loft with a soft blurred
window behind her, lit with Rembrandt lighting and a soft overcast fill, shot
on an 85mm lens with a shallow depth of field, in a photorealistic editorial
look with natural skin texture. Make it a 4:5 portrait frame.

Best for: LinkedIn headshots and magazine-style portraits. See the photorealism guide for skin and lighting deep-dives.

4. Text poster with preset

A retro poster of a lone runner in silhouette on an empty track at dawn,
backlit by golden-hour sun with long shadows, in the retro-poster preset with
a warm limited palette and light grain. Include the ALL-CAPS headline
"NO EXCUSES" in a heavy condensed sans-serif centred near the top. Make it a
4:5 portrait frame.

Best for: gym and event posters — Muse Image renders the quoted headline legibly and the preset sets the era in one word.

5. Reference blend for a brand shot

Combine my two uploads: @ mention photo 1 for the exact sneaker and photo 2
for the model's face. Put the model wearing the sneaker mid-jump on a bright
concrete rooftop at midday, lit by hard midday sun with crisp shadows, shot on
a 35mm lens with a shallow depth of field, in a punchy editorial-film look.
Keep the sneaker and the face exactly consistent. Make it a 4:5 portrait frame.

Best for: social campaigns where the product and the person must stay on-brand — Muse Spark blends both references and holds them consistent.

Frequently Asked Questions

How do I set the aspect ratio in Meta Muse Image?

Pick it from the aspect-ratio control in the Meta AI app or on meta.ai, or state it in words inside your prompt — for example, end with "make it a 9:16 vertical frame." Muse Image supports 1:1 square, 4:3, 3:2, 16:9 landscape, 9:16 vertical, and 4:5 portrait. Because it reads full sentences, spelling the ratio out in plain language works even where no UI toggle is shown, such as WhatsApp.

Does Meta Muse Image use --parameters or flags like Midjourney?

No. Meta Muse Image has no --ar, --style, --v, or any numeric flags. You write full natural-language sentences, not comma-separated tags. Muse Spark, its reasoning layer, plans layout and interprets your prose, so "a wide 16:9 landscape at golden hour" works and "--ar 16:9" does not. Brief it like you would a photographer.

What is Muse Spark?

Muse Spark is the reasoning layer paired with Meta Muse Image. Before it draws, Spark plans the composition and layout, pulls in real-time web context when the prompt needs current facts, blends multiple reference photos, and self-revises its plan to catch errors. It is why a single plain-English sentence produces a coherent, on-brief image instead of a literal keyword mashup.

Can Meta Muse Image render legible text and QR codes?

Yes. Muse Image renders near-perfect legible text, accurate charts and plots, and working, scannable QR codes because Muse Spark writes and executes code behind the scenes to lay them out. Wrap the exact words in quotes, keep headlines short, and name the font. For a QR code, give the exact URL or text it should encode and test the scan before you ship it.

How do I use reference photos in Meta Muse Image?

Upload one or more photos, or @ mention them in your prompt, then say what each is for — blending scenes, keeping a character or product consistent, or transferring a style. For example, "@ mention photo 1 for the person's face and photo 2 for the jacket, and put them in a sunlit park." Muse Spark blends the references and holds the subjects consistent across the result.

How do I edit just one part of an image?

Use region editing — circle, annotate, or sketch over the area you want changed, and Muse Image redraws only that region while leaving the rest untouched. You can also refine conversationally: reply with the single change you want, such as "make the sky overcast," and it keeps everything else identical. Change one thing per message and check the result before the next tweak.

Where can I use Meta Muse Image?

Meta Muse Image is Meta Superintelligence Labs' image model, released in July 2026. You can use it in the Meta AI app, on meta.ai in the browser, inside Instagram Stories in the US, and in WhatsApp. It accepts natural-language prompts and reference photos in all of them; the aspect-ratio and preset controls appear in the app and on meta.ai.

What style presets does Meta Muse Image offer?

Built-in presets include editorial film, Renaissance portrait, retro poster, claymation, 16-bit video game, restore-old-photo, and trending-hairstyle. You can tap a preset or just describe the look in words. Presets are a fast starting point; combining one with your own lighting and lens language in the same sentence gives you the most control.

Advertisement