google-gemini/angular-draw-it-sample ? reverse-engineered prompt

Build me an Angular app that lets someone enter their Gemini API key, type a prompt, and generate images from it. After the images are created, I want to preview them in a nice 3D gallery, then ask questions about the images in a chat style input. The app should use Gemini to understand the images and answer back in plain language, and it should read the answer out loud using the browser’s speech synthesis so it feels interactive. Add a simple instructions area in the UI so a new user can figure out how to use it quickly. Keep it easy to run locally with npm start, and if you need to look up anything current, feel free to check the latest docs online.