Photos of adults who agreed · teams, friends, relatives
AI Group Photo Generator — everyone in one picture
Upload one photo of each person and an AI edit puts them side by side in one scene — same faces, same clothes, one light and a believable height for everyone. For the team that never shares an office and the friends who live on three continents.
- One photo per person
- Up to 16 photos
- Free plan for 2–3 people
- No watermark
Updated by the Kenerate AI team
On this page
Quick answer
How do I make a group photo from separate photos with AI?
To make a group photo with AI, upload one clear photo of each person to a multi-reference image model and describe the shot — "everyone standing together in a park, same light, keep each face the same". Kenerate's AI group photo generator opens the editor with that instruction filled in on Nano Banana 2.1 (up to 14 photos); GPT Image 2.5 Sunburst takes up to 16 and Qwen Image 3.0 does 2–3 people on the free plan. Use photos only of adults who agreed, and say it's an AI image when you share it.
Key facts
- Works on
- One photo of each adult who agreed to be in it
- What it makes
- One new photo with everyone in the same scene and light
- Photos per run
- Up to 3 (Qwen) · 10 (Seedream) · 14 (Nano Banana 2.1) · 16 (GPT Image 2.5 Sunburst)
- Faces hold best
- Small groups — Google lists 4 character images for Nano Banana 2.1
- Not allowed
- Children, public figures, anyone who hasn't agreed
- Watermark
- None
Group size
How many people can be in an AI group photo? Models, photo limits and credits
Credits are for one 1K image, from Kenerate's price calculator. The photo limit is what the editor accepts; faces hold up best in smaller groups — Google's Gemini API docs list up to 4 character images for character consistency on Nano Banana 2.1.
| Group size | Model | Photos it takes | Credits (1K) | Why this one |
|---|---|---|---|---|
| 2–3 people | Qwen Image 3.0 | Up to 3 | 10.8 for 2 · 11.7 for 3 | On the free plan — covers a trio from your starter credits |
| Up to 4 people | Nano Banana 2.1 | Up to 14 | 21, any number of photos | Default here — built for subject consistency |
| 5–10 people | Seedream 5.0 Pro | Up to 10 | 13.5 (27 at 2K), any number | Flat price for a bigger team; run two halves if faces drift |
| Up to 16 people | GPT Image 2.5 Sunburst | Up to 16 | 7.2 + 4.5 per photo (52.2 for 10) | The most photos in one run; spends longer on fine detail |
How to
How to use the AI group photo generator
- Time needed
- Steps
- 4 steps
Step 1
Open the group photo generator
Press “Add the photos”. Image Studio opens on Nano Banana 2.1 with a group-photo instruction in the prompt box. Sign in free for starter credits.
Step 2
Upload one photo per person
Add a clear photo of each adult who agreed to be in the picture — up to 14 on Nano Banana 2.1, 16 on GPT Image 2.5 Sunburst, 3 on the free Qwen Image 3.0.
Step 3
Set the scene and the order
Keep the park at golden hour or write your own: an office, a beach, a living room. Say who stands where by image number, and any heights that matter.
Step 4
Create, check every face, download
Press Create, compare each face with its source photo, rerun if one drifted, and download without a watermark. Label it as an AI image when you share it.
See it in the app

Open Image Studio and type what you want to see — subject, style, lighting and mood.
Watch the tour · 1:23
Prompts, styles, references, edit, Arena & chat
Chapters
Tips
Make the group look photographed together, not pasted
Most fake-looking group photos come from mismatched source photos. Fix those before you press Create.
Use one sharp photo per person, face visible, taken from roughly the same angle — all front-on, or all three-quarter.
Crop every photo the same way: all head-and-shoulders, or all waist-up. Mixing a selfie with a full-length shot breaks the scale.
Say how many people there are and where each goes: “five people; the person from image 1 on the far left, image 2 next to them…”.
Give heights if they matter: “the person from image 3 is the tallest, image 4 is about a head shorter”.
Name one light for the whole scene — “soft overcast daylight” or “warm indoor lamp light” — so nobody looks lit from a different room.
Past four or five people, check every face. If one drifts, run the group in two halves, or rerun and keep the best take.
Models
Models for AI group photos
What each model is good at. Open any one straight in the generator.
Google
Nano Banana 2.1
- Reference images
- Up to 14
- Output
- 1K to 4K
Open Nano Banana 2.1Best for
Default here · built for subject consistency

OpenAI
GPT Image 2.5 Sunburst
- Reference images
- Up to 16
- Output
- 1K to 4K
Open GPT Image 2.5 SunburstBest for
The biggest groups in one run

ByteDance
Seedream 5.0 Pro
- Reference images
- Up to 10
- Output
- Up to 2K
Open Seedream 5.0 ProBest for
Flat price for a team of up to 10
- Free plan

Alibaba
Qwen Image 3.0
- Reference images
- Up to 3
- Output
- 1K or 2K
Open Qwen Image 3.0Best for
Free plan · a trio of friends
FAQ
AI Group Photo Generator questions
10 questions
How do I make a group photo from separate photos with AI?
Upload one clear photo of each person to a multi-reference image model and describe the scene: where everyone stands, the place and the light, and that each face, hairstyle and outfit must stay the same. On Kenerate, press “Add the photos” — Image Studio opens on Nano Banana 2.1 with that instruction filled in — add the photos, press Create and check every face before you download.
Link to this answerIs the AI group photo generator free?
Free to start. A new Kenerate account gets 25 credits on sign-up, and Qwen Image 3.0 is on the free plan: a group of 3 costs 11.7 credits, so the starter credits cover two tries. Nano Banana 2.1 (21 credits at 1K), Seedream 5.0 Pro (13.5) and GPT Image 2.5 Sunburst (7.2 + 4.5 per photo) use credits from one-time packs — there's no subscription. Downloads carry no watermark.
Link to this answerHow many people can I put in one AI group photo?
The editor accepts up to 3 photos on Qwen Image 3.0, 10 on Seedream 5.0 Pro, 14 on Nano Banana 2.1 and 16 on GPT Image 2.5 Sunburst. Accepting a photo isn't the same as keeping that face exact: Google's Gemini API docs list up to 4 character images for character consistency on Nano Banana 2.1. For bigger groups, check every face, rerun when one drifts, or make the group in two halves.
Link to this answerWhy does one person look different in the group photo?
Each extra face is one more thing the model has to hold at once, and image models can drift — OpenAI's own docs say its image model may struggle to keep recurring characters consistent. Use a sharper, front-facing photo of that person, name them by image number (“keep the face from image 3 exactly the same”), shrink the group, or try a different model and keep the best take.
Link to this answerHow do I make the people look the right size and height?
Start with photos cropped the same way — all head-and-shoulders or all waist-up — so the model isn't scaling a close selfie next to a full-length shot. Then write heights in the prompt: “the person from image 2 is the tallest; image 4 is about a head shorter than image 1”. A row of people standing on the same ground, shot at eye level, is the easiest group for a model to get right.
Link to this answerCan I add someone who wasn't there to a real group photo?
Yes, if everyone in it agrees. Upload the original group photo first and the extra person's photo second, and write “add the person from image 2 standing at the right end of the group, same light and scale, keep everyone else exactly the same”. Share it as an edited picture, not as a record of a day that person attended.
Link to this answerCan I make a group photo with children in it?
No. Kenerate is for photos of adults: prompts that mention children, minors or ages under 18 are blocked, and you shouldn't upload photos of anyone under 18. For a family picture, use photos of the adult members who agreed.
Link to this answerCan I make a group photo with a celebrity?
No. Don't put public figures, or anyone who hasn't agreed, into a group photo. A believable picture of a real person somewhere they never were can mislead people — Google's generative AI policy, for one, prohibits impersonating an individual without explicit disclosure in order to deceive. Use photos of people you know who said yes.
Link to this answerWhat group photos can I make with AI?
Team photos for a website or a deck when people work in different cities; friends who live far apart in one holiday picture; relatives who couldn't make the same reunion; a club or band line-up. Write the setting — office lobby, beach, living room, stage — and the light, and keep the same prompt to make matching pictures later.
Link to this answerIs my group photo real or AI — do I have to say?
It's a new AI-made image of real people, so be open about it. Tell everyone in it before you post, and caption it as an AI or edited photo, especially on a company page or anywhere people might take it as a real event. Don't use it to prove that someone was somewhere.
Link to this answerIn depth
AI Group Photo Generator, explained
What works, what to avoid, and how to pick a model.
On this page
How does an AI group photo generator work?
It's a multi-reference edit, not a cut-and-paste collage. All your photos and one instruction go to an image model's edit endpoint together; the model reads each person's face, hair and clothes, then draws a brand-new photo with everyone in the scene you describe — one camera angle, one light, shadows on the same ground.
Because the picture is drawn fresh, the background, poses and even small details can change. That's why the instruction names what must stay (faces, hair, clothes) and why you check each face against its source photo before you download.
How do I choose photos that merge well?
The model can match light and scale far better when the source photos already roughly agree:
- Same framing: all head-and-shoulders, or all waist-up, or all full-length.
- Same angle: everyone front-on or everyone in three-quarter view, camera near eye level.
- Sharp faces: no heavy filters, sunglasses or hands over the face; good daylight is ideal.
- Plain clothes are kept best; busy prints and logos may come back simplified.
- One person per photo. Crop out other people so the model knows who to include.
How do I write a group photo prompt?
Five parts cover almost every group shot:
- Who and how many: “the five people from the five uploaded photos”.
- Where each one stands: by image number, left to right, front or back row.
- The place and the light: “an office lobby with soft window light”, “a beach at sunset”.
- The pose and the camera: “standing side by side, arms around each other, eye level, 35mm lens”.
- What must not change: “keep each person's face, hairstyle, skin tone, build and clothing exactly as in their photo”.
Who can be in an AI group photo?
Only people who know about it and said yes. Three rules keep a group photo fair to everyone in it:
- Content involving minors is blocked: Kenerate blocks prompts about children or anyone under 18, so don't upload their photos.
- Everyone agreed. Ask each person before you upload their photo and again before you post the result.
- No public figures. A believable photo of a real person somewhere they never were can mislead; don't make one.
- Label it. Caption the picture as AI-made or edited, and never use it as proof that something happened.





