Halloween costume photos usually need makeup, props, or an outfit you do not already own, which takes planning ahead of the actual night. This page has seven Gemini prompts that turn a regular photo into a costume portrait, covering classic looks like a vampire, a witch, a zombie, and a skeleton, plus lighter options like a friendly ghost or a cat costume for anyone who wants something less intense. Each prompt describes the makeup, wardrobe, lighting, and background needed for a convincing seasonal photo rather than a flat sticker-like overlay. Copy a prompt, open a Gemini chat, attach a clear photo, and paste the prompt as your message; a sample image next to each prompt shows the target look. Adjusting the intensity is easy: add 'subtle makeup, keep it light' if a look feels too extreme, or swap the 'single red accent light' in the vampire prompt for another colour. Two general tips: a dark or distant photo gives Gemini less of your face to work with, so start with a clear, front-facing, well-lit photo; and if the background looks too busy, for example a cluttered room, reply in the same chat and ask for a 'foggy graveyard backdrop kept simple' or a 'plain dark background,' then check the whole picture again. None of these prompts ask for blood, gore, or anything frightening beyond a normal costume-party level, and every look is described as makeup and wardrobe only, not an actual injury or transformation.
Last tested in Gemini on . How to use these prompts
Prefer not to write prompts? Our sister site AI Halloween Costume turns one selfie into a vampire, witch or zombie photo. The first preview is free, full-size photos are paid, and it is for adults only.
Sample images were made by running each exact prompt in an AI image generator on an AI-generated stand-in photo. Your result will not match the sample exactly: it depends on your photo, the model and version you use, and the randomness of each run.
Only upload a photo of yourself, or of an adult who has agreed to it. Please don't use these prompts on photos of anyone under 18.
AI-generated sample
Restyle the person in the photo as a classic vampire for Halloween: pale skin-tone makeup, dark smoky eyes, a black high-collar cape, subtle fangs, dramatic low-key studio lighting with a single red accent light, and a close-up crop on the face and shoulders.
Copies the prompt; just paste it in Gemini
Tip: Ask for 'no fangs' if you want a subtler vampire look.
AI-generated sample
Turn the person in the photo into a friendly ghost costume: a soft white draped sheet-style costume with simple cut eye holes, bright even indoor lighting, a plain pastel-orange backdrop, cheerful color grading, and a full-body playful pose with arms raised.
Copies the prompt; just paste it in Gemini
Tip: Change 'pastel-orange' to any color that matches your photo shoot.
AI-generated sample
Reimagine the person in the photo as a Halloween witch: a pointed black hat, a flowing dark purple cloak, subtle green-tinted makeup, moody dusk lighting, a foggy outdoor background with bare tree silhouettes, and a half-body pose holding a broomstick.
Copies the prompt; just paste it in Gemini
Tip: Swap 'dark purple' for a different cloak color.
AI-generated sample
Convert the person in the photo into a zombie costume look: greyish pale makeup with subtle faux-torn fabric on the collar, no blood, wounds or scars, tousled messy hair, low dramatic side lighting, a plain dark grey backdrop, desaturated color grading, and a close-up crop on the face.
Copies the prompt; just paste it in Gemini
AI-generated sample
Style the person in the photo with Halloween skeleton face paint: white-and-black bone-pattern makeup, a plain black top, bright even studio lighting, a plain white backdrop, high-contrast black-and-white styling, and a centered close-up headshot.
Copies the prompt; just paste it in Gemini
AI-generated sample
Make the person in the photo wear a playful cat costume for Halloween: small black cat ears headband, painted-on whisker makeup, a black turtleneck, soft warm indoor lighting, a plain cream backdrop, and a half-body pose with a paw-like hand gesture.
Copies the prompt; just paste it in Gemini
AI-generated sample
Place the person in the photo in a foggy Halloween graveyard scene: a simple dark hooded cloak, cool blue moonlit lighting, drifting fog at ground level, silhouettes of plain stone gravestones in the background with no statues, and a full-body pose standing among the fog.
Copies the prompt; just paste it in Gemini
Tip: Add 'less fog' if the background feels too obscured.
Both plain runs of prompt 1 turned the light-brown hair in the photo dark brown or nearly black, and one also turned the grey-green eyes red-brown. With 'Keep the hair colour, hairstyle and face shape exactly as in the photo' added (the first version below), both of our runs kept the light-brown hair, though all four vampire results, these two included, made the face look a little slimmer.
In both of our plain runs, prompt 2's sheet covered the head and body, with only the eyes (through two holes), the arms and the feet showing, so nobody could tell who it was; our sample image, made in ChatGPT, does show the face. Asking for 'a soft white draped sheet worn like a hooded cape with the face uncovered' (the second version below) showed the face in both of our runs, though both also added a smile.
Prompt 6 doesn't mention expression, yet both plain runs smiled, one of them showing teeth. With 'Keep the exact facial expression and mouth from the photo; do not add a smile' added (the third version below), both of our runs kept the neutral expression.
Restyle the person in the photo as a classic vampire for Halloween: pale skin-tone makeup, dark smoky eyes, a black high-collar cape, subtle fangs, dramatic low-key studio lighting with a single red accent light, and a close-up crop on the face and shoulders. Keep the hair colour, hairstyle and face shape exactly as in the photo.
Both of our runs with this line kept the light-brown hair from the photo, though the face still came out a little slimmer; both plain runs of prompt 1 darkened the hair.
Turn the person in the photo into a friendly ghost costume with their face still showing: a soft white draped sheet worn like a hooded cape with the face uncovered, bright even indoor lighting, a plain pastel-orange backdrop, cheerful color grading, and a full-body playful pose with arms raised.
Both of our runs showed the face in a hood-like opening, and both added a smile; both plain runs of prompt 2 hid the face.
Make the person in the photo wear a playful cat costume for Halloween: small black cat ears headband, painted-on whisker makeup, a black turtleneck, soft warm indoor lighting, a plain cream backdrop, and a half-body pose with a paw-like hand gesture. Keep the exact facial expression and mouth from the photo; do not add a smile.
Both of our runs with this line kept a neutral expression; both plain runs of prompt 6 smiled.
: We ran all seven prompts on gemini.google.com in new temporary chats with the model left on Flash-Lite and an AI-generated stand-in photo of a man with light-brown hair; re-ran prompts 1, 2 and 6 as they are; ran each fix twice; ran prompt 3 again with one same-chat follow-up; and, after adding 'no blood, wounds or scars' to prompt 4 and 'no statues' to prompt 7, ran both twice more. None of the twenty-two results showed blood or gore, and no headstone had readable words.
Start with a clear, front-facing, well-lit photo. Heavy makeup effects are easier for Gemini to blend naturally onto real facial features when the source photo is sharp and evenly lit.
Please don't. Every prompt here is written for a photo of an adult, and none of the prompts on this site should be used on photos of anyone under 18. For a child's costume, use the looks as ideas for real makeup and clothes instead.
None of the prompts ask for blood, wounds or gore, and none of our twenty-two Gemini runs for this page showed any; Gemini's zombie came out as a grey, washed-out portrait with a torn collar. Our first zombie sample, made in ChatGPT, did add fake wound makeup, so we added 'no blood, wounds or scars' to prompt 4 and drew a new one.
Yes. In our single test we sent prompt 3, then replied 'Keep everything the same, but add drifting fog at ground level and stone gravestones in the background', and Gemini added both, plus a small tomb we hadn't asked for, and left the rest largely as it was. One test is no guarantee, so check the whole picture again.
