PromptCanvas

Image generation tools need more than a single API call bolted onto a form. PromptCanvas is a full-stack creative tool that turns a text prompt into a generated image via DALL-E, wrapped in a React (MUI) frontend and a Flask backend, and the goal from the start was live editing (regenerate and refine without losing context), proper auth so users own their generated content, and CRUD operations that treat generated images as first-class managed data rather than throwaway outputs.

What I Built

• Text-to-image generation — user prompts routed through the DALL-E API
• Live editing loop — adjust a prompt and regenerate without restarting the flow
• User authentication — registration and login gating access to generation and history
• CRUD APIs for creating, retrieving, updating, and deleting generated images
• MUI-based gallery UI — clean, responsive interface for prompts and past generations
• Flask backend orchestrating OpenAI API calls, rate handling, and persistence

Tech Stack

• Frontend: React, MUI
• Backend: Flask (Python)
• AI: OpenAI DALL-E
• Auth: JWT / session-based

Key Decisions

• Wrapped DALL-E behind a Flask backend rather than calling the API directly from the frontend, so prompt handling, rate limiting, and auth stayed server-side and API keys never touched the client
• Built CRUD APIs around generated images as persisted, ownable resources instead of ephemeral outputs, since users expected to revisit and manage past generations, not just view a single result and lose it
• Supported live regeneration from an edited prompt without resetting the whole flow, since the iterative "tweak and regenerate" loop is how people actually use image generation tools

Outcome / Impact

A production-ready creative tool that demonstrates full-stack AI integration with real auth and data management — not just a thin wrapper around an API call.