This is a submission for the Google AI Studio Multimodal Challenge
What I Built
*Who’s That Pokémon? *– The Twist! is a fun and challenging Pokémon quiz game built with React, TypeScript, and the Google Gemini API.
Can you guess the Pokémon from its silhouette? Careful—the silhouette might not be the Pokémon you think it is!
Demo
GitHub repo: WhosThatPokemon repo
Google AI studio: WhosThatPokemon
Image:
How I Used Google AI Studio
I leveraged Google Gemini API with multimodal inputs to create the core twist mechanic:
- gemini-2.5-flash-image-preview → for morphing Pokémon, reshaping one into another’s silhouette.
- gemini-2.5-flash → for progressive AI clue generation.
The app sends multiple modalities (text instructions + Pokémon images) to Gemini:
- Text Prompt → Explains the morphing rules (e.g., “Take source Pokémon’s colors/texture and apply to silhouette shape”).
- Image 1 → Source Pokémon (for colors and texture).
- Image 2 → Target Pokémon silhouette (for shape).
Output: Gemini processes them together to generate a unique AI morphed Pokémon—a perfect example of multimodality in action.
Multimodal Features
- Image + Text Fusion → Combines text prompts with Pokémon images for morphing.
- AI-Generated Silhouette Morphing → Creates unpredictable twists in gameplay.
- Progressive AI Clue System → AI adapts hints based on player performance.
- Downloadable Artwork → Save and share AI-generated Pokémon creations.
Team
Yusup Almadani
Github : https://github.com/splmdny
Website : https://splmdny.vercel.app/
Top comments (0)