OPEN SOURCEFlutterFirebaseRiverpodiOS + AndroidMIT OPEN SOURCE

Prompt Canvas

An open-source Flutter app that turns text prompts into AI-generated art — with style controls, a community explore feed, and a full Firebase backend.

Prompt Canvas — AI image studio
App screen
App screen
ROLE
Solo Designer & Developer
DURATION
Self-directed build
PLATFORM
iOS & Android
STATUS
Open Source on GitHub
01 — Challenge

Most text-to-image tutorials stop at a single API call — I wanted a complete, well-architected reference app instead.

The goal was a full cross-platform Flutter app that treats AI generation as a real product surface: multi-mode auth, a community explore feed, a personal album, profile management, theming, and onboarding — not a one-screen demo.

The task: structure it cleanly enough to open-source as a reference — a pluggable AI image provider, a Firebase backend, and an architecture (Riverpod, go_router, freezed, fpdart) other developers can read, fork, and ship their own version from.

02 — Process
PHASE 01

Onboarding & auth

Built the animated splash, onboarding flow, and a complete Firebase auth system — email/password, Google Sign-In, and anonymous guest mode — so users can start generating immediately, with a guest path that needs no commitment.

PHASE 02

Generation engine & gallery

Wired a pluggable AI image API (default Vyro.ai Imagine) behind art style, aspect ratio, and quality controls. Every generated image persists to the user's personal album via Firebase Storage and Firestore, with a profile image counter.

PHASE 03

Explore & open source

Added an explore feed to browse community images by style with search, plus light/dark theming (dark-neon default). Released the full source under MIT so others can drop in their own provider keys and run it.

03 — Screens
Home — prompt input with recent generations
Home — prompt input with recent generations
Generate — prompt, image style, and size controls
Generate — prompt, image style, and size controls
Explore — browse community art with search
Explore — browse community art with search
04 — Features
GENERATION
Style-controlled image gen
Pick an art style, aspect ratio, and quality — the app hands the prompt to a pluggable AI image API (default Vyro.ai Imagine), then saves the result to the user's gallery with regenerate built in.
COMMUNITY
Explore feed with search
Browse community-generated images by art style, search for prompts, and discover creations from other users — a social layer built into the core experience.
AUTH
Three-mode authentication
Email/password, Google Sign-In, and anonymous guest mode via Firebase Anonymous Auth — users choose how much to commit before generating their first image.
ARCHITECTURE
Riverpod + go_router + fpdart
Hooks-based Riverpod state, go_router navigation, freezed models, and Either-based error handling (fpdart) — structured cleanly as an open-source reference under MIT.
05 — Results
Open
Source (MIT)
3
Auth methods
Firebase
Full backend
Pluggable
AI provider
NEXT PROJECT
FLUTTER · FIREBASE · IOS + ANDROID
Walnut — Mindful Dating App
A dating app that leads with emotional wellness — live on iOS & Google Play.
↗