Al-Toke FC: AI-generated cinematic portraits for a football team
An AI app that turns player photos into professional cinematic portraits. A 3-stage pipeline with Gemini and Seedream 4 used by a real football squad.
The challenge
Al-Toke FC is a local football team that plays every weekend. They wanted to stand out on social media with professional images of their players — epic shots in the style of top leagues. The problem: hiring a professional photographer and editor for every match simply wasn’t viable.
The mobile phone pictures they were posting weren’t getting any traction. They needed something that would turn ordinary match photos into content people actually wanted to share.
The solution
I built a web application with a 3-stage AI pipeline:
- Classification with Gemini: each uploaded image is analyzed to identify the player, composition and lighting.
- Automated prompt enhancement: the system generates an optimized prompt for the generator based on the classification.
- Cinematic generation with Seedream 4: Google’s model produces the final portrait with the right cinematic style.
Everything runs from a simple mobile-friendly interface. The user uploads a photo, waits a few seconds, and gets back a ready-to-post image.
Technical stack
- Backend: Python with FastAPI
- AI models: Gemini (classification) + Seedream 4 (generation)
- Infrastructure: Google Cloud Run
- Frontend: Vanilla JS + Jinja2, optimized for mobile
The outcome
The team uses the tool every matchday. Players get their professional portrait in seconds, and the club posts them during the game itself. Engagement on these posts is three times higher than on regular photos — likes, comments and shares have all grown.
The system is maintained by the coach, who has no technical knowledge. There’s nothing to configure: you open the app, upload and download.
Result
The team generates cinematic portraits of their players in seconds, straight from mobile. Social media posts tripled their engagement compared to regular photos.