Your selfie as a bouquet of roses with your face in every flower: the bouquet alone, or you holding it. One photo, 38 credits, and the flaws are shown.
Real workflow: a woman in a white T-shirt against a beige wall, then a pastel rose bouquet in kraft paper with her face in each flower, and the same woman holding a white and yellow bouquet against a green wall. The first has ten roses instead of nine, the second eleven.
At a glance
A bouquet of roses, and in the middle of each rose, your head. According to the tracker viralvideoai.app, the idea has been going round since late October 2026. It's our only source and it gives no numbers. We picked it anyway because the image takes one selfie and looks like nothing else in the catalogue.
This workflow gives you two images: the bouquet alone in kraft paper, and you holding it against a green wall. The result is strange, and that's on purpose. Nano Banana Pro carries the demo's face over well but it can't count. We asked for nine flowers and got ten, then eleven, and the faces look placed in the flower, with a hard edge along the forehead.
For a picture of you on a page with other keepsakes, see the scrapbook collage. To see yourself as a toy, photo to figurine, and for something funnier, the caricature. Both renders cost 38 credits, see pricing.
These are the files of the workflow above, exactly as the models produced them, unretouched. The starting point is shown when there is one.
Starting point
Render
People who want to redo the face bouquet trend from a selfie for TikTok or Instagram, content creators looking for a strange image that stops the scroll, friends preparing a joke gift.
A sharp, well-lit selfie facing the camera, of you or someone who agrees, and a description of their face in the prompt. The photo is optional.
Pick a sharp selfie, facing the camera
Your face ends up tiny inside the flower. A blurry photo, a profile or sunglasses leave the model nothing to work from.
Put your features in the prompts
The portrait prompt mentions freckles and a cream sweater. Describe your own face and clothes there, then change the flower colours if you like.
Run the bouquet alone first
Each node costs 19 credits. The bouquet alone shows right away whether the face holds up at small size, so you don't pay for the portrait for nothing.
Count the flowers and zoom in
The model often adds a flower and sometimes leaves a rose with no face. If that bothers you, run it again or crop. Check the foreheads too: that's where the join shows.
The same canvas, tuned for different needs: each variation is two or three lines of prompt away.
The bouquet alone shows the faces better, at nearly twice the size in the demo. The portrait tells more of a story since you're holding your own heads, but the flowers are small. For a thumbnail, take the bouquet alone.
The prompt only says flowers and the model picked roses both times. Write peonies, sunflowers or daisies if you prefer. A sunflower's centre leaves more room for a face. Not tested on the demo.
In the trend, the bouquet repeats a single face. With several photos connected to the node, you can ask for a different face per flower, as long as everyone agrees. The more people there are, the more the faces may start to look alike. Not tested here.
The cream background keeps attention on the bouquet, but you can set it on a kitchen table or in a vase. Change only the sentence about the setting, and take "no vase" out of the prompt.
The prompt leaves it open and the model closed nearly every eye on the bouquet alone, which looks sleepy. Write "all eyes open, looking at the camera" for more presence, even if the result gets more unsettling.
Start from the bouquet alone with a Video node and ask for a slow camera move in with petals trembling. That's the form shown on viralvideoai.app. The demo didn't test this step.
At flower size, the face gets very few pixels. A dark photo gives heads that all look alike and no longer look like you.
"Exactly nine" wasn't enough on the demo. If your idea depends on a precise number, nine relatives for example, count before you post.
A face in a rose can amuse or frighten, depending on the person. Look at the image at full size and think about who's getting it before you send it.
Because image models count badly. The prompt says "exactly nine" and the demo came out with ten, then eleven roses. Another run may land on nine, with no guarantee.
The trend mostly runs as video, with the camera slowly moving in. Connect a Video node to the bouquet for that. The demo stops at the two images, and the video costs extra credits.
Yes if that person agrees, never a famous person or someone who didn't ask for it. A face multiplied across flowers can make people uneasy fast.
38 credits, 19 per render on Nano Banana Pro. Running one image again costs 19 credits.
Yes, by switching the node to Nano Banana 2.1 at 1K, 7 credits a render. The face will be less faithful, and at this size there isn't much margin. The demo runs on Nano Banana Pro.
1792 x 2400 pixels vertical, enough for a story or a small print. On the portrait, each face in a flower is only about a hundred pixels tall.
We don't know. viralvideoai.app only says the idea has been around since late October 2026 and names no creator. Imaginode has no link with that site.
The photo is optional in the canvas, but without one the model invents a woman. For the bouquet to carry your face, it needs your selfie.
Make a flower bouquet of your own face with AI
Use this workflowFree account, no card required. The workflow opens pre-filled in your canvas.
Nano Banana Pro
Flux Kontext
Seedream 5 Pro