Enlarge images without the blur
Stretch a small image in a browser and it goes soft, because the browser just blends neighbouring pixels. This tool gives you two better ways to enlarge at 2× or 4×: Lanczos3 for photos, and an AI model that redraws edges for logos, drawings and screenshots.
Want the best result? Read How to upscale an image without blur.
Lanczos3 or AI: which one to pick
Pick Lanczos3 for photographs. It works out each new pixel from six neighbours on each axis instead of the browser's two, so edges stay noticeably crisper, and it never makes anything up. What was in the photo is what you get, only larger.
Pick AI for logos, icons, line art, cartoons, screenshots and anything with text. The model (Real-ESRGAN) was trained to redraw clean edges, so small lettering and outlines come back sharp instead of fuzzy. On a photo that same habit can make skin and fabric look smooth and painted, which is why it isn't the default.
What the AI method needs
It runs on your graphics card through WebGPU, which current versions of Chrome and Edge support. If your browser doesn't have it, the tool says so and uses Lanczos3 instead of leaving you waiting.
The first time you use it, your browser downloads the model and its runtime, about 36 MB. After that they're kept in the browser's cache, so later images start straight away, even offline. Images up to 1 megapixel (1000 × 1000 pixels, for example) can use AI; anything larger uses Lanczos3.
Two sharpness settings for Lanczos3
Strong is the raw Lanczos3 output with nothing added. It's the best choice when you want maximum detail from a photo.
Mild adds a light unsharp mask that lifts edge contrast without creating halos. It suits images that will mostly be viewed on screen, where a touch of softness looks more natural.
Nothing is uploaded
Both methods run inside your browser tab, the AI one on your own graphics card. The image is decoded, enlarged and re-encoded on your device, with no server involved and no account needed.
Quick questions
Should I use AI or Lanczos3?
Lanczos3 for photos, AI for logos, drawings, screenshots and text. AI redraws edges, which is great for flat graphics but can make a photo look painted.
Is 2× or 4× better?
2× is usually enough for screens and the web. 4× is for print, or when you need to crop into the enlargement. The bigger the jump, the more the tool has to fill in.
Why is the AI option not working in my browser?
It needs WebGPU, which current Chrome and Edge have. Without it the tool tells you and uses Lanczos3, so you still get an enlarged image.
How big is the AI download?
About 36 MB, once. It downloads the first time you use the AI method and is then kept in your browser's cache.
Does this work on any image format?
Any format the browser can decode: JPG, PNG, WebP, AVIF, BMP, GIF and HEIC. Output is JPG, PNG or WebP.
Are my images uploaded?
No. Everything runs in your browser, and the AI method runs on your own graphics card.
Converters
Start here if your files still need converting before you use this tool.