Quick access

AI Image Upscaler

Make small or blurry photos bigger and sharper with AI. Upscale 2× or 4× to as much as 16 megapixels, or enhance a photo at the same size to clean up blur, noise and JPG blocks. Free, no sign-up, no watermark, and your photos never leave your device.

Upscale

The first image downloads a free AI model of about 4.6 MB, which your browser then keeps for next time. Small images take seconds; a large one can take a few minutes on a phone or older computer.

Full size, free

Download the result at up to 16 megapixels (for example 4096 × 4096) as PNG, JPG or WebP, with no credits, watermark or sign-up.

Real detail, not just bigger

The AI model redraws edges, text and textures, and removes blur, noise and JPG blocks that a normal resize only enlarges.

See the difference

Drag the before/after split and zoom in to compare, then set the strength if the result looks too smooth.

Original image

Images

Upscale

Lower it if skin, hair or grass look too smooth or painted; it mixes in a normal enlargement.

Download

Upscaling uses the open Real-ESRGAN model (BSD licence, © Xintao Wang), run in your browser with ONNX Runtime. Licence

Next: shrink the big file with the Image Compressor, cut out the person with the Background Remover, or use the photo in a poster or thumbnail in the Design Studio.

About the AI Image Upscaler

AI Image Upscaler makes small, blurry or low-quality images bigger and sharper. Choose 2× or 4× to make each side two or four times longer, up to 16 megapixels, or Enhance to keep the size and clean up blur, noise and JPG blocks. Unlike a normal resize, which only stretches the pixels, the Real-ESRGAN AI model redraws edges, text and textures, so old phone photos, screenshots, product pictures and images saved from chats come out crisp. Drag the before/after split and zoom in to compare, and lower the AI strength if faces or grass look too smooth. The model runs inside your browser, so your images are never uploaded and there are no credits to buy.

How to use it

  1. 1 Choose 2×, 4× or Enhance, then drop in one or more images, or paste one with Ctrl + V.
  2. 2 Wait while the image is upscaled; the progress bar shows how long is left. Drag the before/after split and zoom in to check the detail.
  3. 3 Set the AI strength and the format, then download the image, or all of them as a ZIP.

Frequently asked questions

Is this image upscaler really free?

Yes. There are no credits, no sign-up and no watermark, and you download the full-size result. Most AI upscalers charge because they run the model on their servers; here it runs on your own device, so it costs nothing to run.

Are my images uploaded to a server?

No. The AI model is downloaded to your browser and runs on your computer or phone, so your images never leave your device. Only your settings, such as the scale and format, are remembered in this browser.

How big can the upscaled image be?

Up to about 16.8 megapixels, such as 4096 × 4096 or 5016 × 3344 pixels, the largest image every browser can safely hold. With 4× that means images up to about 1024 × 1024 pixels; larger images are scaled down first so the result fits, and the page tells you when that happens. Choose 2× for bigger images.

What does Enhance do?

Enhance keeps the image at the same size but runs it through the AI model, which cleans up blur, noise and the blocky marks of heavy JPG compression. It is useful for photos saved from WhatsApp or Facebook, or screenshots that look soft.

Can it restore old or very blurry photos?

It helps a lot with small, soft or compressed photos, but it cannot invent detail that is completely missing. Faces in tiny group photos may look smooth or painted; lower the AI strength to keep more of the original texture.

Why does it take a while?

The image is upscaled on your own device, in small tiles. On a computer with a recent Chrome or Edge, the graphics card (WebGPU) does it in seconds; on the processor alone a 1000-pixel image can take a few minutes. The progress bar shows the time left, and you can stop at any time.

Which model does it use?

Real-ESRGAN general x4v3, an open AI model by Xintao Wang released under the BSD licence, made for real-world photos. It runs in your browser with ONNX Runtime Web.