AI Image Prompts & Model GuideGenerated on 2026-10-18 · 8 Real 1:1 Outputs

Halloween AI Photo Prompts: 8 Tested Ideas & Real Outputs

Prices updated 2026-10-18

Below are 8 copy-ready Halloween AI photo prompts covering classic candlelit portraits, golden-hour pumpkin patches, Victorian Gothic interiors, retro 1950s house parties, and studio beauty makeup. Every card pairs the exact English prompt we ran with a real 1:1 sample image (displayed at 768×768 px on this page) generated on 2026-10-18 through UniRoute using Nano Banana 2 ($0.05 per 1K, 1:1 image) or GPT Image 2.5 Flare ($0.01875 per 1K, 1:1 image).

Disclosure: UniRoute is our own unified AI API gateway. The 8 samples (shown at 768×768 px on this page) were generated through UniRoute on 2026-10-18; prices are from UniRoute’s media catalog on that date.

How to Use These Prompts

Step 1

Copy a Prompt

Choose a card below, copy its prompt, and adapt the details using the customization tip.

Step 2

Open the Model Page

Open the model listed on the card to review its inputs, settings, and current price.

Step 3

Review Price & Generate

Generating on UniRoute requires a signed-in account with an active balance (there is no free tier; each image is billed at the model’s catalog price).

Note on photo uploads & reference faces: Photo upload is not supported on UniRoute today (all image models in our catalog take text-only input). If you use another tool that accepts a reference photo, you can try adding this optional line (not tested by us): Use the person in the attached photo and keep their face unchanged. Only use photos of yourself or people who have agreed.

8 Copy-Ready Prompts

Each card lists its model, settings, and sample output to help you choose and adapt a prompt.

Witch, Pumpkin Patch & Victorian Mansion Portraits (Prompts 1–3)

Candlelit black-velvet witch close-ups, golden-hour pumpkin-patch portraits, and haunted Victorian staircase scenes.

AI-generated classic witch portrait beside glowing jack-o-lanterns at dusk, generated with Nano Banana 2 on UniRoute
#classic-witch-portrait1K, 1:1 · $0.05

1. Classic Candlelit Witch Portrait

Nano Banana 2
Prompt (75 words)

Editorial close-up portrait of an adult woman wearing a wide-brimmed black velvet witch hat, dark plum cloak, and subtle silver crescent jewelry, standing beside carved glowing jack-o'-lanterns on a mossy stone porch at dusk. Warm flickering candlelight balances against cool twilight mist in the background. Shot on a full-frame camera with an 85mm f/1.8 lens, shallow depth of field, crisp skin texture, rich autumn color grading, photorealistic studio quality, no text, no watermark, no signage.

Warm candlelit 85mm close-up with velvet textures, glowing carved pumpkins, and cool twilight mist.

How to customize: Swap "dark plum cloak" for emerald or burgundy velvet, or change "85mm f/1.8" to "35mm f/1.8" to show more of the stone porch.

Model & option: nano-banana-2 (1K, 1:1)Catalog price (2026-10-18): $0.05
AI-generated golden-hour autumn pumpkin patch portrait, generated with Nano Banana 2 on UniRoute
#pumpkin-patch-autumn-portrait1K, 1:1 · $0.05

2. Golden-Hour Pumpkin Patch Portrait

Nano Banana 2
Prompt (72 words)

Natural lifestyle portrait of an adult woman in a burnt-orange knit sweater, plaid wool scarf, and dark felt fedora, crouching among heirloom orange and white pumpkins in a rustic autumn pumpkin patch. Golden-hour sunset backlight filters through amber oak leaves with soft rim light on her hair. Captured on a 50mm f/1.4 prime lens, creamy bokeh, warm autumnal film tones, inviting seasonal atmosphere, photorealistic editorial photography, no text, no watermark, no signage.

Daylight autumn lifestyle portrait with golden-hour rim light, knitwear textures, and creamy 50mm bokeh.

How to customize: Replace the sweater and scarf with a trench coat or denim jacket, or switch "sunset backlight" to "overcast misty morning light" for a moodier palette.

Model & option: nano-banana-2 (1K, 1:1)Catalog price (2026-10-18): $0.05
AI-generated Victorian manor staircase portrait with candlelit chiaroscuro lighting, generated with Nano Banana 2 on UniRoute
#haunted-victorian-mansion-portrait1K, 1:1 · $0.05

3. Haunted Victorian Manor Staircase Portrait

Nano Banana 2
Prompt (74 words)

Dramatic medium portrait of an adult man dressed in a tailored charcoal Victorian frock coat, brocade vest, and antique brass pocket watch chain, standing on a carved wooden staircase inside an old candlelit manor house. Moody chiaroscuro lighting from brass wall sconces casts long shadows across weathered wallpaper and drifting dust motes. Shot on a 35mm f/1.8 lens, cinematic Gothic romance mood, fine fabric detail, photorealistic portrait photography, no text, no watermark, no signage.

Cinematic 35mm environmental portrait in a candlelit Victorian staircase with dramatic chiaroscuro shadows.

How to customize: Change the subject to an adult woman in a Victorian velvet gown, or adjust "35mm f/1.8" to "50mm f/1.4" for a tighter upper-body framing.

Model & option: nano-banana-2 (1K, 1:1)Catalog price (2026-10-18): $0.05

Retro Party, Gothic Vampire & Skeleton Glam Looks (Prompts 4–6)

Warm 1950s living-room masquerade snapshots, candlelit stone-hall Gothic vampire portraits, and couture skeleton face-paint studio portraits.

AI-generated retro 1950s Halloween party photograph with two adult friends in masquerade masks, generated with Nano Banana 2 on UniRoute
#retro-1950s-halloween-party1K, 1:1 · $0.05

4. Retro 1950s Halloween House Party

Nano Banana 2
Prompt (71 words)

Candid vintage-style photograph of two adult friends laughing at a 1950s living-room Halloween party, wearing classic mid-century masquerade masks, polka-dot swing dress, and tweed suit with paper streamers and carved pumpkins on a wooden sideboard. Warm tungsten table lamps mixed with gentle vintage camera flash create nostalgic mid-century color contrast. Shot on a 35mm lens, subtle film grain, joyful retro party atmosphere, photorealistic editorial photography, no text, no watermark, no signage.

Nostalgic mid-century two-person party scene combining warm tungsten lamps, vintage flash, and subtle film grain.

How to customize: Swap "1950s living-room" for "1970s autumn parlor" or change the outfits to classic vintage cat-eye masks and velvet blazers.

Model & option: nano-banana-2 (1K, 1:1)Catalog price (2026-10-18): $0.05
AI-generated Gothic vampire portrait in a candlelit stone hall, generated with GPT Image 2.5 Flare on UniRoute
#vampire-gothic-portrait1K, 1:1 · $0.01875

5. Gothic Crimson Velvet Vampire Portrait

GPT Image 2.5 Flare
Prompt (73 words)

High-fashion Gothic portrait of an adult man wearing a deep crimson velvet collar cape, black silk cravat, and silver signet ring, posed in a vaulted stone hall lined with tall beeswax candelabras. Cool blue moonlight streams through a lancet window while warm candlelight sculpts dramatic cheekbones and dark eyes. Captured with an 85mm f/1.4 portrait lens, shallow depth of field, elegant nocturnal mood, photorealistic magazine editorial style, no text, no watermark, no signage.

Dual-color temperature 85mm Gothic editorial portrait mixing cool lancet-window moonlight with warm candelabra key light.

How to customize: Adjust "adult man" to "adult woman in a crimson velvet corset gown" or replace "vaulted stone hall" with "misty wrought-iron courtyard at midnight".

Model & option: gpt-image-2-5-flare-text-to-image (1K, 1:1)Catalog price (2026-10-18): $0.01875
AI-generated couture skeleton face-paint beauty studio portrait, generated with GPT Image 2.5 Flare on UniRoute
#skeleton-face-paint-glam1K, 1:1 · $0.01875

6. Couture Skeleton Face-Paint Beauty Portrait

GPT Image 2.5 Flare
Prompt (73 words)

Beauty studio portrait of an adult woman wearing intricate monochrome couture skeleton face paint with metallic silver eyeliner, black lace high-neck gown, and dark floral crown against a charcoal backdrop. Controlled softbox key light and crisp rim lighting highlight fine brushstroke pigment texture and shimmering eye makeup. Shot on a 100mm f/2.8 macro portrait lens, tack-sharp focus, luxury fashion magazine aesthetic, refined glamorous Halloween mood, photorealistic photography, no text, no watermark, no signage.

100mm macro beauty studio portrait highlighting metallic silver eyeliner, brushwork face-paint texture, and black lace.

How to customize: Change "metallic silver eyeliner" to "antique gold leaf accents" or add "marigold flower crown" for a richer autumn color pop.

Model & option: gpt-image-2-5-flare-text-to-image (1K, 1:1)Catalog price (2026-10-18): $0.01875

Cozy Rainy-Window & Cinematic Foggy Forest Looks (Prompts 7–8)

Cozy rainy-window evenings with friends and spiced cider, and atmospheric twilight foggy-forest lantern portraits.

AI-generated cozy Halloween evening lifestyle photo by a rainy window, generated with GPT Image 2.5 Flare on UniRoute
#cozy-halloween-at-home1K, 1:1 · $0.01875

7. Cozy Rainy-Window Halloween Evening at Home

GPT Image 2.5 Flare
Prompt (74 words)

Warm lifestyle photograph of two adult friends in oversized autumn knit sweaters and witch headband accessories sitting on a plush window seat with ceramic mugs of spiced cider, surrounded by mini white pumpkins and fairy string lights. Soft amber lamp glow contrasts against rainy blue evening glass outside. Shot on a 35mm f/1.4 lens, shallow depth of field, cozy hygge autumn atmosphere, natural candid expressions, photorealistic indoor photography, no text, no watermark, no signage.

Cozy indoor 35mm two-person lifestyle photo pairing warm fairy lights and spiced cider with a rainy blue window.

How to customize: Change "two adult friends" to "solo adult reader holding an open hardcover book" if you prefer a single-subject cozy portrait.

Model & option: gpt-image-2-5-flare-text-to-image (1K, 1:1)Catalog price (2026-10-18): $0.01875
AI-generated cinematic foggy forest portrait with a glowing brass lantern, generated with GPT Image 2.5 Flare on UniRoute
#cinematic-foggy-forest-lantern1K, 1:1 · $0.01875

8. Cinematic Foggy Forest Lantern Portrait

GPT Image 2.5 Flare
Prompt (73 words)

Cinematic environmental portrait of an adult traveler wearing a hooded dark wool cloak and leather satchel, holding a glowing vintage brass lantern along a leaf-strewn forest trail wrapped in low evening fog. Golden lantern light illuminates swirling mist and mossy tree trunks against deep teal twilight shadows. Shot on a 50mm f/1.4 lens, volumetric light rays, atmospheric autumn mystery mood, rich dynamic range, photorealistic cinematic still photography, no text, no watermark, no signage.

Atmospheric 50mm twilight forest portrait with volumetric brass lantern light cutting through teal woodland mist.

How to customize: Use option "1K:16:9" when generating widescreen desktop wallpapers or video storyboards from this foggy forest scene.

Model & option: gpt-image-2-5-flare-text-to-image (1K, 1:1)Catalog price (2026-10-18): $0.01875

The 7-Part Structure Used in These Prompts

When adapting these prompts for different outfits, props, or settings, keep this 7-slot order intact and swap one or two visual nouns at a time:

[Photo Style & Framing] + [Fictional Adult Subject] + [Wardrobe, Fabric & Makeup] + [Seasonal Setting & Props] + [Key Light vs. Ambient Color Temperature] + [Camera Focal Length & Aperture] + [no text, no watermark, no signage]

1. Pair Warm and Cool Light Sources

Instead of vague mood words, pair a warm key light (flickering candlelight, brass lantern, amber lamp) against a cool ambient fill (twilight mist, blue moonlight, rainy evening glass).

2. Match Focal Length to Framing

Prompts 1, 5, and 6 use 85mm or 100mm lenses for closer portrait and makeup framing; prompts 2, 3, 4, 7, and 8 use 35mm or 50mm lenses to include the pumpkin patch, staircase, party room, or forest trail.

3. Describe Fabrics Instead of Franchise Characters

We avoid trademarked and copyrighted movie characters and instead name the exact fabrics, garments, and makeup details—such as black velvet, a brocade vest, a crimson collar cape, or metallic silver eyeliner.

UniRoute Image Models & Catalog Pricing (As of 2026-10-18)

Catalog options and per-image prices from UniRoute’s live media catalog (GET /v1/media/catalog) on 2026-10-18. All five models take text-only input:

Model / Catalog IDProviderBase OptionsBase Price (USD)Other Tiers / OptionsSample Usage & Notes
Nano Banana 2
nano-banana-2
Google1K:1:1, 1K:16:9$0.052K:1:1 / 2K:16:9 ($0.075), 4K:1:1 / 4K:16:9 ($0.1125)Used for samples 1–4 at 1K:1:1 ($0.05/image). Text-only input; offers 1:1 and 16:9 at 1K, 2K, and 4K.
GPT Image 2.5 Flare
gpt-image-2-5-flare-text-to-image
OpenAI1K:1:1, 1K:16:9$0.018752K:1:1 / 2K:16:9 ($0.0625), 4K:1:1 / 4K:16:9 ($0.1)Used for samples 5–8 at 1K:1:1 ($0.01875/image). Text-only input; offers 1:1 and 16:9 at 1K, 2K, and 4K.
Nano Banana Pro
nano-banana-pro
Google1:1$0.11251:1 only ($0.1125)Not used for these samples. Text-only input; offers 1:1 at $0.1125 per image.
Google Nano Banana
google/nano-banana
Google1:1$0.018751:1 only ($0.01875)Not used for these samples. Text-only input; offers 1:1 at $0.01875 per image.
Seedream 4.5 Text to Image
seedream/4.5-text-to-image
ByteDance1:1, 16:9, 9:16$0.0406251:1, 16:9, 9:16 ($0.040625)Not used for these samples. Text-only input; offers 1:1, 16:9, and 9:16 at $0.040625 per image.

Developer Quickstart: Run These Prompts via UniRoute’s Media Image API

The samples on this page were generated with the UniRoute Media Image API (/docs/media-image). Following the Media API quickstart (/docs/media-quickstart), store your key in a private file (such as ~/.config/uniroute/api-key) and pass it via a temporary header file (-H @"$AUTH_HEADER") to keep the raw key off the command line: quote first with POST /v1/media/quote, replace <unique-id> with a fresh unique ID per request in the Idempotency-Key header, create the task with POST /v1/media/tasks using top-level media_type, model, option, prompt, and max_cost_usd fields, then poll GET /v1/media/tasks/{task_id} after HTTP 202 until status is "success" and billing.state is "charged" before downloading from urls[].url:

POST /v1/media/quote → POST /v1/media/tasks → GET /v1/media/tasks/{task_id}
umask 077
KEY_FILE="$HOME/.config/uniroute/api-key"
AUTH_HEADER="$(mktemp)"
BODY="$(mktemp)"
printf 'Authorization: Bearer %s\n' "$(cat "$KEY_FILE")" > "$AUTH_HEADER"
trap 'rm -f "$AUTH_HEADER" "$BODY"' EXIT

# 1) Quote the exact price for model + option
curl -sS -X POST https://api.uniroute.app/v1/media/quote \
  -H @"$AUTH_HEADER" \
  -H "Content-Type: application/json" \
  -d '{"model":"nano-banana-2","option":"1K:1:1"}'

# 2) Write the task JSON body with a quoted heredoc (safe for any prompt text with single quotes)
cat > "$BODY" <<'JSON'
{
  "media_type": "image",
  "model": "nano-banana-2",
  "option": "1K:1:1",
  "prompt": "Editorial close-up portrait of an adult woman wearing a wide-brimmed black velvet witch hat, dark plum cloak, and subtle silver crescent jewelry, standing beside carved glowing jack-o'-lanterns on a mossy stone porch at dusk. Warm flickering candlelight balances against cool twilight mist in the background. Shot on a full-frame camera with an 85mm f/1.8 lens, shallow depth of field, crisp skin texture, rich autumn color grading, photorealistic studio quality, no text, no watermark, no signage.",
  "max_cost_usd": "0.05"
}
JSON

# 3) Submit the image task (returns HTTP 202 with data.task_id)
#    Replace <unique-id> with a fresh unique string per request (e.g. from python3 -c "import uuid; print(uuid.uuid4())")
curl -sS -X POST https://api.uniroute.app/v1/media/tasks \
  -H @"$AUTH_HEADER" \
  -H "Idempotency-Key: <unique-id>" \
  -H "Content-Type: application/json" \
  --data @"$BODY"

# 4) Poll until data.status == "success" and data.billing.state == "charged",
#    then download from data.urls[].url (do not send Authorization to the download URL)
TASK_ID="<task_id from step 3>"
curl -sS https://api.uniroute.app/v1/media/tasks/"$TASK_ID" \
  -H @"$AUTH_HEADER"

Frequently Asked Questions

Do these prompts work in Gemini or Nano Banana?

These are plain-text English prompts. Our 8 sample images on this page were generated on 2026-10-18 using Nano Banana 2 (samples 1–4) and GPT Image 2.5 Flare (samples 5–8) through UniRoute. We did not test them in the consumer Gemini app or other third-party apps.

Can I use my own photo to generate a Halloween portrait?

Not on UniRoute today—every image model in our catalog currently takes text-only input and does not support photo uploads. If you use another tool that accepts reference photos, you can try adding "Use the person in the attached photo and keep their face unchanged" to the prompt (not tested by us). Only use photos of yourself or people who have agreed.

Is it free to generate these Halloween photos on UniRoute?

No. UniRoute does not provide free image credits; generating an image requires a signed-in account with a balance, and each image is charged at the model’s catalog price. As of 2026-10-18, Nano Banana 2 (1K:1:1) is $0.05 per image and GPT Image 2.5 Flare (1K:1:1) is $0.01875 per image.

How much did all 8 Halloween sample photos cost to generate?

On 2026-10-18, generating prompts 1–4 with Nano Banana 2 (1K:1:1) cost $0.05 × 4 = $0.20, and generating prompts 5–8 with GPT Image 2.5 Flare (1K:1:1) cost $0.01875 × 4 = $0.075—a total of $0.275 for all 8 sample images.

Can I change the aspect ratio from 1:1 to 16:9 or 9:16?

As of 2026-10-18 on UniRoute, Nano Banana 2 and GPT Image 2.5 Flare offer 1:1 and 16:9 options at 1K, 2K, and 4K (such as "1K:1:1" or "1K:16:9"); neither offers a 9:16 option. Nano Banana Pro and Google Nano Banana offer 1:1 only. Seedream 4.5 Text to Image offers 1:1, 16:9, and 9:16 ($0.040625 per image; not used for these samples).

Can I make a video from these Halloween prompts?

The prompts on this page were written and tested for still image generation only. To check available video models, options, and request parameters on UniRoute, see the video API documentation at /docs/media-video.

Read Video API Documentation (/docs/media-video)

Why do these prompts avoid famous horror movie character names?

We avoid trademarked and copyrighted characters and instead describe wardrobe, fabrics, props, and lighting directly—such as a black velvet witch hat, a crimson velvet collar cape, or monochrome couture skeleton face paint.

Ready to Create with These Prompts?

Explore the models used on this page, read the Media API quickstart, or check current catalog pricing: