Next.js frontend gives Gemini image generation a dark-mode facelift

What it does This is a Next.js 14 web frontend for AI image generation. You type a prompt and get back images, or upload an existing image and instruct the model to edit it via natural-language chat. It supports both Google’s Gemini 2.5 Flash Image Preview and ByteDance’s Doubao SeedReam 4.0, and can generate one to four images at a time in styles like anime, realistic, or enhanced detail.

The interesting bit The project is essentially a well-dressed API client—what the README bills as “exquisite” is a dark-themed, gradient-heavy UI with hover animations and loading spinners. The practical value is that it bundles two different Chinese and Western image APIs behind a single interface, so you can switch models without changing tools.

Key highlights

• Built with Next.js 14 and TypeScript, deployable on Vercel.
• Supports both Gemini 2.5 Flash and Doubao SeedReam 4.0 backends.
• Offers text-to-image and conversational image editing (upload + natural language instruction).
• Generates 1–4 images per prompt with selectable styles: enhanced, artistic, anime, or realistic.
• Responsive dark-mode design with gradient cards and loading animations.

Caveats

• Requires separate API keys for both Gemini and Doubao, plus a custom Doubao API base URL; it is not a self-hosted model.
• The Doubao integration relies on a third-party endpoint (apipro.maynor1024.live) rather than ByteDance’s official console directly, which may raise stability or compliance questions the README does not address.

Verdict Worth a look if you want a ready-made, mobile-friendly interface for Gemini 2.5 Flash Image Preview or need to compare outputs against Doubao without writing your own client. Skip it if you are looking for a local, open-source diffusion model that runs offline.