Show HN: Open-source Veo/Imagen social app (Flutter FastAPI mock mode) (github.com)

An open-source social app demo that plugs Google’s Veo 3.1 (video) and Imagen 4.0 (image) Vertex AI models into a Flutter front end and FastAPI backend has been released. Users build a one-time “AI avatar” by uploading three profile views (front, left, right) and can then generate personalized images or up-to-8-second videos—“me skateboarding through cyberpunk Tokyo”—by toggling “Include Me” and adding up to three per-post reference images. The repo includes a smart feed (Interest 60% / Explore 25% / Trending 15%), real-time Firebase updates, multi-aspect outputs (1:1, 9:16, 16:9), private/public controls, and client/server mock mode for local development. For developers the stack is ready-made for experimentation: Flutter (Dart) UI, FastAPI (Python 3.11+) backend, Vertex AI for generation, Firestore and Firebase Storage for data, and Riverpod state management. Production use requires a GCP account and budget (estimated ~$0.10–0.30 per video, $0.02–0.05 per image; $10–50+/mo depending on usage), while ENABLE_MOCKS=true simulates generation for free testing. This project is significant because it lowers the barrier to prototyping identity-aware generative experiences, but also highlights practical concerns—cost management, content moderation, rate limiting, privacy and TOS—making it a useful reference for builders exploring personalized generative media.