What's actually happening in these images

A single AI-generated photo shows a subject — a real person from an uploaded photo, a pet, or a fictional character — rendered as a plastic collectible action figure, still sealed in toy-aisle blister-pack packaging: a printed cardboard backing card with a character name and faux-brand logo, a clear plastic bubble, and a handful of small accessory props laid out next to the figure that reference the subject's actual hobbies, job, or personality. It's built on Google Gemini's image-generation model — nicknamed "Nano Banana" by users during an early testing period, a name that stuck in casual use even after the feature shipped under Gemini's regular branding. The appeal is personalization: two people's figures look nothing alike because the accessories, packaging text, and pose are all drawn from their specific input.

Not the same as the tiny-world rescue trend

Worth clarifying up front, since the two get mentioned together constantly: this is a single static image, made with an AI image generator. Our tiny-world rescue trend is a short video, made with an AI video generator (typically Kling). Some creators chain them — generate an action figure image first, then animate it into a miniature-scene video on a separate tool — but they're two different outputs built on two different kinds of model, not one trend with two names.

The prompt structure

The format is popular partly because it's a genuinely repeatable recipe:

  1. Reference the subject. Upload a clear photo (person, pet, or describe a character) as the image the figure should resemble.
  2. Specify the toy format. "Collectible action figure in blister-pack packaging" — the packaging is what sells the illusion, not just the figure itself.
  3. Describe the packaging details. A printed name/title on the cardboard backing, a faux brand logo, toy-photography studio lighting with visible plastic sheen on the figure.
  4. List 3-5 accessories. Small props tied to the subject's real interests, job, or personality — a coffee cup and laptop for a remote worker, a tiny leash and ball for a dog, a guitar for a musician. This is the step that makes each result feel personal instead of generic.

Being specific in steps 3 and 4 is what separates a figure that looks like a real toy-aisle product photo from one that just looks like "a photo with a plastic filter." Vague prompts produce generic results; naming the exact accessories and packaging text produces something worth sharing.

Which tool to use

Google's Gemini app (or Google AI Studio, for more control over the prompt) is the tool this trend is actually built on — it's not a generic prompt that works equally well on any image generator, since the "Nano Banana" nickname refers to this specific model's particular strength at this kind of stylized, packaging-aware composite. Other image generators can attempt a similar look, but Gemini is the one credited across creator tutorials for this exact format.

Once you've got a figure you like

The output is one flat image — figure, packaging, and background all baked together with no separate layers. If you want just the figure by itself (for a profile picture, a sticker, or to composite onto a different background), that's a background-removal step done after generation. RemoveGifBG handles static images as well as GIFs and video — upload the PNG or JPG and get the figure back on a transparent background. See why transparent backgrounds matter for when you actually need one.

Try RemoveGifBG Free

Related: Tiny-world rescue trend · AI pet portrait trend · How AI videos are made

FAQ

What is the Nano Banana AI action figure trend?

It's a viral image trend that turns a photo of a person, pet, or character into a static image of a collectible-toy-style action figure — complete with blister-pack packaging, a printed character name, and a handful of tiny plastic "accessories" laid out beside it, the way a real toy-aisle figure is photographed. It's built on Google Gemini's image-generation model, nicknamed "Nano Banana" by users during its testing period, and the name stuck even after the feature shipped under Gemini's normal branding.

Is this the same as the tiny-world rescue video trend?

No. The action-figure trend produces a single static image using an AI image generator (Gemini/Nano Banana). The tiny-world rescue trend produces a short video using an AI video generator (typically Kling). Some creators combine them — generate an action figure first, then animate it into a miniature-world video on a separate tool — but they're two different outputs from two different processes.

What's the actual prompt for the action figure trend?

The repeatable structure: describe a blister-pack action figure of the subject in the uploaded photo, specify the toy-photography style (studio lighting, plastic sheen, a printed cardboard backing with a character name and logo), and list 3-5 small accessory items relevant to the subject's personality or hobbies laid out next to the figure. Being specific about the packaging text and accessory choices is what makes each result feel personalized rather than generic.

Can I get a transparent cutout of my AI action figure?

The generated image comes as one flat picture — the figure, its packaging, and the background are all baked into a single image with no separate layers. If you want just the figure (or just the packaging) on its own to use as a sticker, profile picture, or overlay, that requires a background-removal pass after generation. Upload the image to a background remover and export the subject with a transparent background.