How to resize an image without losing quality
"Resize without losing quality" means two different things depending on direction. Making an image smaller is almost free. Making it larger cannot add detail that was never captured, and every tool that claims otherwise is guessing.
Smaller is safe
Scaling a 4000 pixel photo down to 1200 throws away pixels, but the picture was carrying far more detail than a screen at that size can show. Done properly, the smaller image looks as good as the original at that size, and often sharper.
"Done properly" is the catch. Scaling down by simply skipping pixels produces aliasing: jagged diagonal edges, and shimmering moiré patterns on fabric, brickwork and screens. A good resizer averages groups of source pixels into each output pixel instead. Our image resizer asks the browser for its high-quality smoothing, which averages rather than skips.
Bigger is guessing
A 600 pixel image contains 600 pixels of information across. Scale it to 2400 and each original pixel has to become sixteen. The resizer has to invent the fifteen new ones, and there are only two ways to do it.
- Interpolation (bilinear, bicubic, Lanczos): blend the neighbouring pixels. Fast, honest, and soft. Edges blur because the tool is averaging across them. Fine for up to about 1.5×; beyond 2× it looks noticeably out of focus.
- AI upscaling: a model that has seen millions of images predicts plausible detail — hair texture, brick edges, skin. At 2× and 4× it can look dramatically better than interpolation. The image upscaler offers 2× and 4×.
AI upscaling is guessing, just better-looking guessing. It is right for a print of a small holiday photo. It is wrong for anything where the detail has to be true: text you need to read accurately, a licence plate, a document, evidence. Those pixels are invented, however convincing they look.
The only way to genuinely get a bigger image is a bigger source. If there is any chance of finding the original file, that beats every upscaler.
The order of operations
Most lost quality comes from doing reasonable things in the wrong order.
- Start from the original. Not a copy downloaded from a chat app or a social site; those have already been shrunk and recompressed.
- Crop first. Cropping removes pixels you do not need, so the resize works on less and nothing is scaled twice. Use the cropper.
- Resize once, directly to the final size. Resizing to 2000, then 1500, then 1080 blurs a little at each step.
- Sharpen lightly, after resizing, never before. Downscaling softens edges slightly; a small unsharp mask brings them back. Use a small radius and stop before halos appear around edges. The sharpener has radius and threshold controls for exactly this.
- Save once, in the right format, at a sensible quality. For JPEG and WebP photos, 80–85 is plenty.
Why a resized image can be bigger in file size
A common surprise: you resize a JPEG to fewer pixels and the file gets larger. That almost always means it was saved in a different format or at a higher quality than it came in at. A photo saved as PNG is routinely five times the size of the same photo as JPEG. Keep photos as JPEG or WebP, and if you need to hit a size limit, run the result through the image compressor.
Many files at once
If you have a folder of product photos or holiday pictures to bring to one size, the bulk image resizer applies the same settings to all of them and gives you a zip. The same rules apply: start from originals, and resize once.