How to Make AI Images of Yourself (Portraits, Styles, Scenes)

The quickest way to make AI images of yourself is to upload a clear selfie to an image tool that accepts photos, then describe the new picture and say what it should keep from your face. ChatGPT, the Gemini app and Toybox AI's Image Generator all work this way. Below are the steps in each, the photos that give the closest likeness, five prompts to copy, the consent rules for other people's faces, and fixes for a face that comes out wrong.

5 ways to make AI images of yourself

There are two kinds of tools: general image makers, where you describe the whole picture, and single-job tools built around one kind of result, such as a headshot. As of September 2026:

A reference photo is a picture you upload so the AI can use it as a guide for a face, a style or a layout. The result is a brand-new picture rather than an edit of your photo, which is why the face can shift. The page on using a reference image with an AI image generator covers what a photo can and can't steer.

Which photos of yourself to upload

Upload a photo that shows your face plainly. When a selfie is dim, angled or filtered, the AI has to guess at your features, and a guessed face can look like a stranger.

• One person, face toward the camera. Crop out friends and strangers so there's no question which face is yours.
• Daylight on your face. Stand facing a window so your eyes and jawline are lit, not shaded by an overhead lamp.
• No beauty filter. A filter reshapes your jaw and smooths your skin, so the photo no longer shows your real face.
• Your everyday look. Wear the glasses you always wear and your current hairstyle, so the result looks like the person your friends know.
• Eyes visible. Skip sunglasses and hats pulled low.
• The original file. Upload the photo itself, not a screenshot of it.
• A second angle, if the tool takes one. Add a three-quarter view as a second reference photo, then compare the likeness with and without it.
• A file the tool accepts. Toybox takes JPG or PNG photos up to 10 MB. If your iPhone saves photos as HEIF, switch new photos to JPEG: open Settings, tap Camera, then Formats, and choose Most Compatible.

How to make AI images of yourself with ChatGPT

ChatGPT Images is available on every plan, on the web and in the iOS and Android apps. OpenAI says ChatGPT Images 2.5, whose system card was published September 8, 2026, makes the subjects of your reference photos more recognizable. Free accounts have a separate usage limit for creating images, and ChatGPT tells you when you reach it.

1. Start a chat, or pick Images from the sidebar.
2. Upload your photo in the message box.
3. Describe the new picture: the setting, your clothes, the style, and what to keep, for example "Keep my face, hair and glasses exactly as in my photo."
4. Choose a shape with the aspect ratio picker, or write it in the prompt, such as 4:5 for an Instagram post.
5. Send it. OpenAI says a complex request can take a few minutes, and you can keep chatting while it works.
6. To change one area, open the image in the editor, choose Select, highlight the area and describe the change. OpenAI warns that highlights aren't always precise and edits can reach past the area you picked.
7. Select Save to download the picture. ChatGPT also keeps every image you make under Images.

Shared prompts are another way in. OpenAI's launch example was an '80s flashback prompt that shows you as you might have looked in that decade, and anyone who opens a shared prompt adds their own photo and details. For more looks built this way, browse these AI photo trend prompts or the prompts for an '80s photo of yourself.

How to make pictures of yourself in the Gemini app

Google's help page names this exact use for the Gemini app: seeing yourself with different hairstyles or in different scenes. With a personal Google Account you must be 18 or older to edit images, and a school account can have other limits.

1. Sign in on the Gemini website or in the Gemini app.
2. Upload your photo. You can upload several images and ask Gemini to build one new picture from them.
3. Start the request with a verb such as "create" and name the style, for example "Create a photorealistic image of me with a short bob haircut, keeping my face the same."
4. Download the image. Without a Google AI plan, downloads are 1K; with a plan, 2K.

Gemini can also keep an avatar of you. You record your face and voice once with your phone, then type @ and your Google username in a prompt instead of adding a selfie to every request. As of September 2026, making one takes a personal Google Account and an age of 18 or older, and avatars can't be used in the UK, Switzerland or the EEA.

In the US, adults can link their Google Photos library to Gemini. Gemini then uses the face groups you've named in Google Photos to make images of those people, so name only the groups of people who would be fine with it. Google's help page also reminds users not to violate other people's copyright or privacy.

Prompts for AI images of yourself

These prompts are written for Image Generator. Add your photo as a reference photo before you create the image, and swap in your own details. You can paste them into ChatGPT or Gemini too, though each tool reads them differently.

A trail photo that should look like a real snapshot:

Use my reference photo for my face, hair and glasses. Show me hiking a mountain trail at golden hour in a red rain jacket and a gray backpack, looking back over my shoulder with a relaxed smile. Shot on a phone at eye level, natural skin texture, soft haze in the valley behind me. Keep my face shape, eye color and hairline the same.

A pencil portrait for a gift or a profile picture:

Draw the person in my reference photo as a graphite pencil portrait on warm gray toned paper, head and shoulders, with white chalk highlights on the cheekbones and hair. Keep my face shape, nose, eyebrows and hairstyle recognizable. Plain background, no text.

A vintage travel poster in a flat illustrated style:

A 1960s travel poster illustration of the person in my reference photo walking along a seaside boardwalk, flat colors in teal, coral and cream, bold shapes, a striped umbrella and gulls in the sky. Keep my hairstyle, glasses and face shape. No words on the poster.

You and your cat, using two reference photos:

Use the first reference photo for me and the second for my cat. Show me curled up in a window seat with a paperback on a rainy afternoon, the cat asleep on my lap, a knitted blanket over my knees and rain on the glass behind us. Soft gray daylight, photo style. Keep my face and hair the same, and keep the cat's coat pattern and eye color.

Two people, for a picture the other person has agreed to:

Use the first reference photo for me and the second for my sister. Show the two of us laughing at a kitchen table while we make fresh pasta, flour on our hands, window light from the left, photo style at eye level. Keep each face and hairstyle as it is in its own photo, and keep my sister's glasses.

Say which photo is whom whenever there are two people. Without that, features from the two faces can end up mixed.

Can AI generate images of real people?

Yes. ChatGPT, the Gemini app and Toybox's Image Generator can all make a picture of a real person from an uploaded photo, so the limits on whose face you use come from each tool's rules, its safety checks and the law. Before you make AI images of people other than you, know these rules:

• OpenAI's usage policies (updated October 29, 2025) forbid using someone's likeness, such as a photorealistic picture or their voice, without consent when the result could mislead people about what's real.
• Google's Generative AI Prohibited Use Policy bans impersonating a real person, living or dead, to deceive people without clearly saying so, and bans content that facilitates non-consensual intimate imagery.
• Toybox's Terms of Service prohibit generating images of real people without their explicit consent.
• Intimate images are off-limits. Never make or share a sexual or intimate image of a real person. If someone posts one of you, report it to the platform right away and keep a screenshot of where it appeared.

In practice: ask each adult before you upload their photo, ask a parent before you use a child's, and show people the result before it goes online. Never make an image that puts a real person somewhere they weren't in a way others could believe, such as a public figure at an event.

Troubleshooting

Put yourself in a new scene with Image Generator

Image Generator in Toybox makes one new picture from your description, guided by up to 3 reference photos of you, your pet or a person who agreed. Toybox AI works in your browser on a phone or computer, with no download.

1. Open Image Generator and sign in with email or Google. The first time, verify your email.
2. Type the scene into "Describe your image", and name what to keep: face shape, hair, glasses.
3. Tap "Add reference photos" and pick up to 3 JPG or PNG photos.
4. Pick a "Size". Square 1:1 suits a profile picture, and Story 9:16 fills a phone screen for Reels.
5. Tap "Create image". Most image tools finish in about 30 seconds.
6. Hold the result next to your selfie and check the eyes, nose and hairline. On a Pro plan, type a fix into "Want to change something?" (currently 85 credits), such as "make my glasses round". On the Free plan, change your description and create a new image.

Toybox doesn't promise an exact likeness, because the AI treats your photos as a guide rather than copying them. Each image currently costs 50 credits, about 50 cents at top-up prices, and the 20 free credits new accounts get don't cover one (see current prices). For a work photo, Portrait Pro starts from one selfie and offers Looks such as Corporate, LinkedIn and Studio; the guide to turning a selfie into a headshot explains how to take the photo you upload. To see yourself in aviators, stubble or a cowboy hat before you buy or grow anything, Glow Up Studio adds the accessories you pick, with "How bold?" set to Natural, Extra or Chaos. Toy Studio turns the same selfie into a collectible toy shot, such as a boxed action figure or a plushie.