How to do an AI photoshoot from your selfies

Steps

1. 1

   Open the AI Photoshoot

   Go to the AI Photoshoot tool page. The studio there is set up for a photoshoot, and the prompt is written for you from the settings you pick. You can go through them before you sign in.

2. 2

   Upload your photos

   Under Your photos, add 1 to 4 clear photos of the same person, with the face visible, good light and no sunglasses. Nano Banana, the default model, takes up to 3; the other models take 4. If you are not signed in, a sign-in box opens when you upload: create an account with an email and password, or continue with Google.

3. 3

   Choose a look

   Each look is a complete setting, outfit and light, such as a white studio backdrop with soft key light or golden hour on a city street. Pick one for a single image.

4. 4

   Add details (optional)

   If the page has an optional text field, such as an outfit, type it there, for example "black blazer over a white tee". It applies to every look. Leave it empty and that part is dropped from the prompt.

5. 5

   Pick a model and shape

   Nano Banana is selected by default. The menu also has Nano Banana 2, Nano Banana Pro, Seedream 5.0 Lite, Seedream 4.5, GPT Image 2 and FLUX.2 [pro]. Portrait 3:4 is preselected; 1:1, 4:3, 16:9 and 9:16 are also available. Seedream 5.0 Lite, Seedream 4.5 and GPT Image 2 can make up to 4 images in one run.

6. 6

   Press Generate, or run a Full shoot

   Generate makes one image in the look you picked; the button shows the credit price before anything is spent, including the number of images you picked. Full shoot makes one image per look with the same photos and model, and its button shows the total (the number of looks times the price per image) before you spend anything. Each look that fails is refunded on its own, and the shots are grouped as one shoot in your account.

7. 7

   Download or share

   Download the shots you like, or share one to the public gallery. Nothing is shared unless you choose to.

Tips

• Use photos with even light and your whole face visible. Sunglasses, hats and heavy filters make your likeness harder to keep.
• Two to four photos from different angles give the model more to work with than one.
• Run the same look on two models. Nano Banana keeps faces close, Seedream adds more cinematic color.
• Check the likeness on a single image before you run a Full shoot on every look.
• Switch to 9:16 for stories and reels, or 1:1 for a profile picture.
• Open "Prompt sent to the model" under the settings to see the exact prompt your choices produce.

Troubleshooting

The face does not look like me

Add more photos (up to the model's limit) with clear, even light, and generate again. Nano Banana Pro is sharper if the default is not close enough.

I cannot add a fourth photo

Nano Banana takes up to 3 reference photos. Switch to another model in the menu to use 4.

The message starts with "This request was blocked"

Our own check stopped the request, often because of something typed in a text field. No credits were used. The link in the message explains the rule.

The message says "The model's safety filter rejected this request"

The model provider refused the request. Credits are refunded up to a daily limit. Rephrase what you typed or try another model.

The upload says "File is too large" or "Unsupported file type"

Use a PNG, JPEG or WebP under the size limit shown in the message. Export a smaller JPEG and upload it again.

The generation failed

The credits are refunded automatically and the result card says so. Try again or pick another model.

Questions