If you've ever spent twenty minutes hunting for the right prompt to get Nano Banana or GPT Image to output exactly the poster style you had in mind, this post — and the tool I built — is for you.
I recently launched HereImg, a site that combines two things I personally kept needing and never found bundled together:
- A searchable gallery of AI image prompts (organized by model and category, not just a random Twitter thread)
- A set of fast, free, browser-based image editing tools for the boring-but-essential stuff — resize, crop, compress, remove background, etc.
Here's why I built it, what's under the hood, and how it might save you some time.
The problem: prompt engineering for images is still trial-and-error
Text-to-image models like Nano Banana, GPT Image, and Seedream are incredible, but getting consistent, high-quality output still comes down to prompt phrasing. Most people either:
- Scroll through scattered social media posts hoping to find a prompt that matches their use case, or
- Burn through credits re-rolling generations with slightly different wording
I wanted a place where prompts are organized like a real reference library — by model (Nano Banana / GPT Image / Seedream) and by category (Food & Drink, Illustration & 3D, Photography, Poster Design, Product & Brand, UI & Graphic) — so you can browse examples close to what you're trying to make, copy a working prompt, and tweak it instead of starting from zero.
Every prompt in the gallery is paired with its actual generated output, so you're not guessing what a phrase like "cinematic rim lighting, 35mm" will actually produce.
The other half: tools you need before and after generating
AI generation is rarely the last step. You still need to resize the output for Instagram, crop it for a thumbnail, strip the background for a product shot, or compress it before uploading. So I bundled a Magic Tools suite alongside the prompt gallery:
- Social Media Resizer (pre-set dimensions for major platforms)
- Image Cropper / Resizer / Rotator & Flip
- Compressor & Format Converter (JPG/PNG/WebP)
- Brightness & Contrast, Filters, Grayscale
- Color Picker (extract hex codes / palettes)
- Watermark tool
- AI Background Remover
The key design decision: all the "free tools" run entirely client-side in the browser. Nothing gets uploaded to a server. That means no signup, no file size anxiety, no waiting on an upload queue — and your images never leave your device. Only the AI-powered features (like background removal) touch a backend, and those are clearly marked with a credit cost.
Under the hood
For fellow builders curious about the stack: HereImg runs on Cloudflare Workers, which has been a great fit for this kind of product —
- Edge deployment means the tool feels instant no matter where in the world someone opens it
- Cheap to run at scale, which matters a lot as an indie/bootstrapped project
- Pairs naturally with client-side image processing, since the heavy lifting (canvas/WASM operations) happens in the visitor's browser rather than on my infra
If there's interest, I can write a follow-up post specifically on the Workers + D1 setup, the KV caching strategy for the prompt gallery, and some of the deployment gotchas I hit along the way (happy to share the pain so you don't repeat it).
Try it out
- 🖼️ Browse the prompt gallery
- 🛠️ Magic Tools suite
- 🎁 New accounts get 20 free credits to try AI generation/background removal
It's free to use, no login required for the browser-based tools, and I'd genuinely love feedback from this community — especially if you spot a prompt category I'm missing or a tool you wish existed.
If you build something with it (or find a prompt that works great), drop it in the comments — I'll add good ones to the gallery.
Top comments (0)