About this tool
Make a small or low-resolution image bigger and crisper. A super-resolution neural network (Real-ESRGAN's compact x4 model) runs directly in your browser with WebAssembly, reconstructing detail that a plain resize can't — cleaner edges, less blur, reduced JPEG noise. It works best on small images (icons, avatars, old photos, product shots). Nothing is uploaded: the AI model downloads once to your device and every image is processed locally, so it's free, private, and there are no limits. Because the model runs on your CPU, large images are processed in tiles and can take a little while.
Frequently asked questions
Is my image uploaded?
No. The AI model is downloaded to your browser once (about 15 MB including the runtime) and runs on your device. Your image never leaves your computer — no server, no account, no limits.
Why does it take a while?
Super-resolution is heavy maths, and it runs on your CPU via WebAssembly rather than a cloud GPU. Small images finish in a few seconds; large ones are split into tiles and can take up to a minute. Pick 2× for a faster, lighter result.
What size can it handle?
Very large inputs are scaled down to 1000 px on the longest side first, then upscaled, so the output tops out around 4000 px. The tool shines on small images — that's exactly what upscaling is for.
2× or 4×?
The model always reconstructs at 4×; the 2× option simply displays and saves that result at half size — smaller file, still enhanced. Both take the same time to compute.