How to Upscale a Low-Resolution Photo 2× or 4× for Printing or Cropping

TL;DR

Upscaling turns a small image into a larger one with AI-added detail, up to a long side of 4096 px. Use 2× for a modest boost and 4× for tiny images. Use Quality for important photos and Fast for speed. It helps printing and cropping; it doesn't make a poor photo good.

The problem

An image is too small for what you need: an old download, a tight crop of a bigger shot, a thumbnail someone sent. A normal resize only stretches the existing pixels, so the picture gets bigger and blurrier at the same time. Printing a small image at A4 is a good way to see this.

What upscaling does

Instead of stretching, a super-resolution model synthesises plausible detail as it enlarges: edges become cleaner and textures reappear. PhotoRepair does it in overlapping tiles on the device, with no upload.

  • 2× or 4× enlargement
  • Quality (Real-ESRGAN) or Fast (OmniSR)
  • The result’s long side is limited to about 4096 px

Steps

  1. Open Enhance and choose your photo.
  2. Pick the factor and the mode.
  3. Run it and check the size and detail.
  4. Compare faces and any text.
  5. Save at the new size.

When it’s worth it

  • Printing a small image larger
  • Cropping into a detail and keeping some sharpness
  • Reviving old downloads and heavily compressed images
  • Preparing an image for a bigger screen

When it isn’t

  • The photo is already large. The 4096 px cap will limit the enlargement.
  • The photo is badly blurred. See unblur.
  • It’s just for a social post. The platform will shrink it again.

Tips

  • Start with the cleanest source you have; the model amplifies what’s there, including noise.
  • Check faces. If they look plasticky, try 2× or Fast.
  • For old family prints that are also faded or damaged, see restoring old photos.

FAQ

What is the maximum size? The result is limited to about 4096 px on the long side to keep memory in check, so already-large photos get a smaller effective enlargement.

2× or 4×? 2× for a natural, modest boost; 4× for very small images. 4× can look smoother and less natural.

Quality or Fast? Quality uses Real-ESRGAN for the best result. Fast uses a lighter model for quicker results.