Halloween AI Photo Prompts: 10 Costume & Makeup Looks
By Chris Morgan
Every October, the same problem: the costume idea is better than what the camera catches. This year the fix is a Halloween AI photo prompt. Paste one of the ten below into ChatGPT or Gemini, attach a clear photo of yourself where the look calls for a face, and you get a preview of the full look — makeup, lighting, mood — before you spend a dollar on fake blood.
Quick answer: copy a prompt, fill in any [BRACKETED] slot, and when you want your own face in the result, add this line at the very front: keep the exact same face, skin tone, and features from the photo I upload. Do not change my identity. Each example on this page was made from the prompt shown with it, so what you see is what the text actually produces.
Prefer one workspace? Every prompt on this page also runs inside Oimi AI Studio Canvas — 30+ top image models including GPT-Image-2.5 and Nano Banana 2 in one place, free to try. Paste the prompt, drop in your photo, and see which model keeps your face best under the makeup.
How to use these prompts
- Pick your lane. The first five looks are makeup and costume transformations — they want your face. The next three are full scenes that work with or without you. The last two are illustration and lifestyle shots for cards and decor.
- Upload the right photo. Front-facing, even light, no sunglasses, face filling a good part of the frame. Halloween looks replace a lot — but the face underneath still decides whether the result looks like you.
- Fill the brackets, keep the scaffolding. Slots like [COLOR] or [FABRIC] are where your details go; the words around them are the design system.
- Fix in one line, not a redo. Keep everything the same, only make the mask more reflective beats regenerating and hoping.
One practical note for the season: these looks peak on Instagram in the last two weeks of October. Generating early means you can reuse the best one as your costume brief for an actual party — the prompt doubles as a shopping list.
Halloween makeup & costume transformations
Skull face paint, red hoodie
Selective color noir: everything gray except the hoodie, skull makeup doing the scaring. One of the most striking profile-picture looks of the season.
A close-up portrait of a person wearing a vibrant red hooded sweatshirt, with the rest of the scene desaturated into high-contrast monochrome. The subject features intricate skull face paint with deep black shading around the eyes and jawline, creating a stark, skeletal aesthetic. A spiked wristband adds a gritty, industrial texture to the foreground, drawing focus to the hand resting near the hood. The background wall is illuminated by harsh, directional light filtered through Venetian blinds, casting distinct parallel shadow stripes across the surface. The composition emphasizes the selective color saturation of the red fabric against the gritty, noir-inspired grayscale environment, highlighting the sharp contrast between the organic skin texture and the artificial, theatrical makeup.
* **Style:** Selective color photography, high-contrast noir
* **Key elements:** Skull makeup, red hoodie, spiked wristband, Venetian blind shadows
* **Lighting:** Harsh directional light, high contrast, neutral white balance with selective red saturation
* **Camera:** Close-up, 50mm lens, eye-level angle
Tweak: the whole look hinges on the selective color. If red starts bleeding into the background, add keep everything except the hoodie fully desaturated at the end.
The cybernetic skull
Half SFX makeup, half machine: cracked porcelain skull, mismatched eye sockets, cables wired into the head. The darkest look on this page.
Cybernetic horror portrait, gaunt humanoid figure with cracked porcelain-white skull-like mask, mismatched hollow eye sockets (one sunken void, one recessed metallic ring), jagged exposed teeth, surrounded by a chaotic tangle of thick black cables and industrial bobbin/coil attachments wired into the head, tattered dark fabric top, dramatic low-key lighting, deep black background, high contrast monochrome, horror photography, cinematic, hyperdetailed texture, 85mm lens, shallow depth.
Tweak: if the result drifts toward robot rather than horror, move cracked porcelain-white skull-like mask to the front of the prompt — the model gives weight to what it reads first.
The raven costume
A blouse grown from black raven feathers, dissolving into ink and smoke. This one is written as a structured brief and built for a reference photo — attach your picture and the "reference image" line does the identity work.
{
"image_style": "surreal, dark cinematic, ethereal",
"subject": {
"person": "woman with features of reference image",
"pose": "head tilted back, eyes closed, expression of freedom and release",
"framing": "close-up portrait from chest upward",
"clothing": "long-sleeve blouse made of organic, layered deep matte black raven feathers",
"accessories": "large black hoop earrings"
},
"visual_effects": {
"hair": "loose, dynamically suspended, blowing upward and forward by wind",
"dissolving_elements": [
"feathers and hair merging with black ink splatters",
"charcoal dust",
"wisps of dark smoke"
],
"face_details": "fine black ink splatters and wisps spreading across face and neck"
},
"environment": {
"background": "smoky gray, moody atmosphere",
"elements": "thick, dense, layered smoke swirling throughout the frame"
},
"technical_aspects": {
"resolution": "high-resolution",
"lighting": "cinematic, moody, diffused through smoke"
}
}
Tweak: the JSON format is the point — swap any value and the rest of the structure holds. If the ink splatters swallow your features, change "face_details" to very fine, sparse ink specks.
The masquerade mask
One jeweled mask, one visible eye, deep red everywhere. The most elegant option here, and the one that survives contact with an actual party.
Editorial beauty portrait of a young woman holding a crystal-encrusted red rhinestone masquerade mask over her left eye, one striking blue eye visible, bold red lips, winged eyeliner, soft glowy skin, hair loosely pinned up, dramatic red seamless background, deep shadows casting across the face, sheer red fabric draped over shoulder, high fashion editorial lighting, ultra-sharp details on the jeweled mask, shot on medium format camera, professional studio photography
Tweak: ask for the mask held slightly away from the eye if your version glues it to the face — a held mask reads as a person, a stuck-on mask reads as a sticker.
Charcoal skin, gold eyes
Body paint as art direction: matte black skin, hundreds of water droplets, and metallic gold eyeliner as the only color. Not a costume — a statement.
Vertical 9:16 extreme close-up beauty portrait of a serene human face in three-quarter profile, filling almost the entire frame. The subject's eyes are gently closed, with a calm, meditative expression and slightly parted matte-black lips. The skin is completely coated in deep charcoal-black pigment with a subtle rough, mineral-like texture. Hundreds of realistic glossy water droplets of different sizes cover the forehead, temples, cheeks, nose, eyelids, lips, and jawline. Each droplet is sharply defined, rounded by surface tension, and reflects tiny white studio highlights.
The eyelids are painted in intensely luminous metallic golden-yellow pigment, creating the only vivid color in the composition. The yellow makeup should appear richly saturated, slightly textured, and almost internally glowing, with soft golden light spreading subtly around the eyelids. Preserve realistic eyelashes, eyebrow hairs, pores, lip lines, facial contours, and natural skin anatomy beneath the black body paint.
Use dramatic low-key studio lighting from the upper-left side, producing crisp highlights on the droplets and gentle sculpting across the nose, cheekbones, lips, and jaw. Keep the opposite side of the face partially fading into shadow. Add a faint secondary golden glow around the distant closed eye. The background is pure dark charcoal-black with a smooth gradient and no visible objects.
Macro beauty photography, ultra-sharp foreground details, shallow depth of field, softly blurred rear side of the face, photorealistic water physics, cinematic contrast, minimalist black-and-gold palette, luxurious editorial cosmetics campaign, hypnotic, elegant, mysterious, meditative, high dynamic range, fine-art photography, 100mm macro lens, f/2.8, studio-quality image, extremely detailed, 8K.
Tweak: the droplet count is the effect. If your version comes out with a handful of drops, repeat hundreds of realistic glossy water droplets, sharply defined near the end of the prompt.
Witchy scenes & gothic portraits
The baroque countess
Painted, not photographed: Caravaggio lighting, cracked gold leaf, a spectral figure in a desecrated sanctuary. The most expensive-looking image on this page, and it costs one prompt.
Visual Style:A masterpiece of High-Baroque Tenebrism, reminiscent of a haunting, sacrilegious commission by Caravaggio or Artemisia Gentileschi. The aesthetic is defined by extreme chiaroscuro, where subjects emerge from an absolute, ink-black void into a singular, violent shaft of theatrical light. The texture is hyper-emphasized: thick, buttery impasto strokes create a 3D physical relief on the canvas, featuring heavy craquelure (cracked paint) and visible, aggressive brushwork. The color palette is a "bruised" spectrum—deep oxblood, oxidized copper green, scorched umber, and flecks of crumbling gold leaf that seem to flake off the surface. Every highlight is a sharp, brilliant strike of white-gold light, while every shadow is a dense, suffocating space where detail is swallowed by darkness.Subject Description:A spectral, porcelain-skinned woman sits enthroned in a position of agonizing grace within a desecrated sanctuary. Her hair is a chaotic storm of obsidian curls and silver threads, rendered with such thick paint it appears to have physical weight and movement. She wears a tattered, heavy silk kimono that hangs in jagged, sculptural folds; the fabric is weighed down by dying embroidery of withered peonies and weeping willow branches. Her arms are a canvas of "living" ink—traditional irezumi tattoos that appear to pulse with a dull, subterranean light under her skin. She holds a steady, terrifyingly intimate gaze, her eyes reflecting an ancient, forbidden knowledge. The background is a crumbling altar of cracked stone, where a massive, decaying mandala is carved into the masonry. Suspended in the air are particles of dust and drifting ash, caught in the harsh light like tiny stars, creating a sense of stagnant, holy silence.
Tweak: if the scene gets too dark to read, ask for a wider shaft of light and more visible craquelure — the cracks are what sell the "old painting" illusion.
The cliff-top witch
Storm light, circling ravens, a gown of black feathers on a windswept cliff. Full scene, works with or without your face in it.
A hauntingly beautiful woman stands on a windswept cliff beneath a stormy twilight sky. Her long dark hair flows wildly in the wind as ravens circle above and around her, their wings glinting with faint moonlight. She wears an elegant gown made of layered black feathers and silk, the fabric rippling like smoke. A delicate choker of pearls and obsidian adorns her pale neck, and her gaze is distant, filled with ancient sorrow and quiet power. The ocean crashes below, sending mist through the air, and the faint glow of the setting sun paints her silhouette in silver and shadow. The atmosphere is gothic and ethereal, blending beauty with melancholy. Cinematic composition, dramatic lighting, high detail, soft depth of field, painterly realism.
Tweak: "ancient sorrow and quiet power" is what keeps the expression regal instead of spooky-mascara. Keep that phrase even when you shorten everything else.
The hooded monochrome
The prompt below is a three-variation system; Variation 2 (Studio Hood) is the Halloween one — a narrow beam across the eyes, everything else swallowed by black.
Ultra-realistic fine art black-and-white portrait, editorial fashion photography, close-up centered composition, direct eye contact, calm yet emotionally powerful expression, flawless natural skin texture with subtle freckles and pores, cinematic high-contrast monochrome grading, dramatic Rembrandt lighting with soft directional light, deep blacks and luminous highlights, razor-sharp eyes with catchlights, premium medium-format camera quality, 85mm portrait lens, f/1.8, shallow depth of field, museum-quality fine art photography, minimalist composition, timeless elegance, luxury Vogue-style editorial, rich tonal range, ultra-detailed facial features, photorealistic, HDR, 8K resolution, award-winning portrait photography.
Variation 1 (Underwater): Half-submerged underwater portrait, crystal-clear water surface crossing the forehead, natural water caustic light patterns dancing across the face and shoulders, floating hair, tiny air bubbles, soft underwater haze, blurred fish in the distant background, tranquil atmosphere, surreal yet realistic, monochrome fine-art aesthetic.
Variation 2 (Studio Hood): Black hooded outfit against a pure black background, narrow beam of light illuminating only the eyes and upper face, soft shadow falloff, symmetrical composition, dramatic chiaroscuro, elegant fashion editorial styling, mysterious cinematic mood, luxury magazine cover aesthetic.
Negative Prompt: low resolution, blurry, overprocessed skin, waxy face, plastic texture, extra limbs, deformed eyes, asymmetrical face, duplicated features, noisy image, oversharpened, artifacts, watermark, logo, text, cropped head, bad anatomy, unrealistic lighting, cartoon, CGI, low quality, distorted proportions.
Tweak: copy the whole text and add use Variation 2 on the first line, or paste just the Variation 2 paragraph. If the beam widens too much, repeat narrow beam, only the eyes and upper face.
Cozy & printable
The jack-o'-lantern latte
Foam art, burlap, pine cones, pumpkins in the background — October in one frame. No face needed, which makes it the easiest post of the season.
A hyper-realistic close-up of a black coffee orange cup with frothy cappuccino foam featuring a smiling jack-o'-lantern face made of cocoa powder and chocolate syrup, on a burlap napkin, surrounded by autumn leaves, pine cones, and Halloween pumpkins in the background with warm, soft lighting, extreme detail, ultra HD, sharp focus, rich textures, cinematic atmosphere, style photorealistic
Tweak: the foam design is swappable: a cat face, a bat, or a small ghost all hold just as well as the jack-o'-lantern.
The retro cat witch
A grumpy black cat in a witch hat, drawn in distressed 1970s style. Print it on a card, a tote, or the goodie bags — it scales without falling apart.
A grumpy black cat wearing a tiny witch hat, holding a small orange pumpkin, sitting on a broomstick, vintage retro style, 1970s distressed typography feel, warm orange and black color palette, white background, high quality vector art, cute but sassy expression --no text remove background
Tweak: the --no text flag keeps the design clean for print. Change "warm orange and black" to purple and lime and you have a different decade of Halloween entirely.
FAQ
Can I put my own face in these?
Yes — for the five looks in the costume section, add the identity line from the Quick answer and attach one clear front-facing photo. The three scene prompts take a face gracefully, and the latte and cat-witch prompts don't use a face at all. One rule covers all ten: if the face in your result isn't yours, the problem is almost always the source photo, not the prompt.
ChatGPT or Gemini — which one for Halloween looks?
Both run every prompt on this page, with one exception: number 10 uses Midjourney-style flags (--no text remove background) — on ChatGPT or Gemini, just delete that tail and say no text in plain English instead. ChatGPT holds long structured prompts best — the raven JSON and the baroque countess read like film briefs on purpose. Gemini (Nano Banana) tends to keep your face more faithful through heavy makeup looks like the skull and the charcoal-gold, which is exactly what you want when the look buries your features. Oimi AI Studio Canvas covers all three — Midjourney included, so number 10's flags work as-is — alongside Nano Banana 2 and GPT-Image-2.5, so you can run the same prompt on several models and keep the winner, free to start.
How early is too early to post these?
Instagram historically starts rewarding Halloween content in the second week of October, and the "Add Yours" chains peak around the 25th–31st. Generating in late September means your best results are ready to post the day the window opens — and you can still order props matching whatever look you chose.
Are Halloween character looks a copyright problem?
Generic monsters are free: witches, vampires, skulls, ghosts, ravens. The line is named characters — a prompt asking for a specific movie's likeness is a different situation from "porcelain skull mask with cables," which describes a look, not a franchise. Keep your prompts descriptive and your prints are yours to share.
Can I print these for cards and party decor?
The latte shot and the cat witch are the natural print picks — flat backgrounds and clean composition survive resizing. Ask the model for its highest resolution and keep the original aspect ratio; for the vector cat, requesting a higher resolution version of the same design beats letting it redraw the composition.
Keep reading
- AI Portrait Poster Prompts: 10 Copy-Paste Designs That Look Art-Directed (2026) — turn any portrait into print-quality poster art, from Japanese ink minimalism to glitch and graffiti
- 10 Trending ChatGPT Photo Editing Prompts to Copy (2026) — upload your photo and try trending portrait edits free
- 5 Trending 80s Prompts for ChatGPT (Copy-Paste Ready) — the retro style that went viral on Instagram, ready to copy
- 100+ Awesome Nano Banana Prompts: Copy & Generate Instantly — a living library of prompts across posters, 3D miniatures, and character design
About the author
Chris Morgan writes about AI image tools and the prompt workflows behind the portraits filling Instagram and Reels — what works, what quietly stops working, and why. This page is part of a running archive of looks still worth copying, refreshed as older styles start to date.