This is a submission for the Google AI Studio Multimodal Challenge
What I Built
Pet fashion is everywhere, on Instagram, TikTok, and even magazine covers, but pet owners have no easy way to see how their own animals would look in those same outfits.
I built Pet Wallpaper Creator, an applet that solves this gap by letting owners transfer clothing from another pet’s image directly onto their own pet’s photo. The goal is to turn the question “how would my pet look in that?” into a personalized, high-quality result.
Unlike simple overlays or stickers, this system performs outfit transfer. The clothing is taken from another pet’s image and adapted onto the user’s pet photo. This involves aligning the garment to the animal’s body, accounting for differences in pose and perspective, and blending textures so the final image looks natural while preserving the pet’s unique identity. With pets’ varied body shapes, fur patterns, and postures, the difficulty is even greater.
The user experience is simple but powered by multimodal capabilities under the hood powered by nano banana and Gemini 2.5 Pro
Upload a photo of your pet and a photo of another pet wearing the desired outfit.
The app generates a new portrait where the user’s pet is realistically dressed.
The app analyzes this new image and creates a witty, outfit-aware caption.
The end result can be either downloaded as a new portrait or shared as a social-media-ready post, making it both a creative personalization tool and a demonstration of how multimodal AI can combine image and text generation into one seamless experience.
Demo
Link to Deployed Applet: https://pet-wallpaper-creator-125320442146.us-west1.run.app/
GoogleAI Studio Link: https://ai.studio/apps/drive/1KO9GrIn64pTyHO_DPPyKj6McUMLu8zV9
Screenshots:
The input page showing the pet photo upload, the outfit-from-another-pet upload, and name field.
The generated portrait with the transferred outfit in wallpaper preview.
The full social-post output image (formatted for Instagram etc.) with caption.
Video Walkthrough:
https://youtu.be/RAf3yqcGBgo
How I Used Google AI Studio
I used Google AI Studio to prototype and chain together two multimodal tasks.
With Gemini 2.5 Flash Image (Nano Banana), I refined prompts so an outfit from one pet photo could be transferred realistically onto another, preserving pose and identity.
With Gemini 2.5 Flash Text, I built a captioning step where the generated image was analyzed to produce a playful, outfit-specific caption that included the pet’s name.
AI Studio made it simple to test, iterate, and connect these capabilities into a working pipeline before deployment.
Multimodal Features
The app combines two multimodal features:
Outfit Transfer Between Pets (Image + Image + Text → New Image)
Users upload their pet photo and an outfit photo from another pet. The model aligns and blends them into a natural, high-quality portrait.Outfit-Aware Captioning (Image + Text → New Text)
The generated portrait is passed into a text model that writes a witty, shareable caption tailored to the new clothes.
Top comments (0)