Change image dimensions. When you enlarge, an AI upscaling model restores detail instead of stretching pixels, so the result does not go soft.
How to use it
- Upload one or more images.
- Choose scale, longest edge, or exact pixels, and enter the value.
- When enlarging, pick the upscaling model that matches the picture — photo or illustration.
- Run it and download the results, individually or all at once.
Doesn't enlarging make images blurry?
Plain enlargement does, because it only stretches the pixels that already exist. Enlarging here runs an upscaling model that reconstructs edges and texture, so the result stays sharp. It is a reconstruction, not a recovery — detail that was never captured cannot come back.
Upscaling is slow.
It runs on your device, so speed depends on your machine. Large images are processed in tiles, and the progress indicator shows how many tiles are done. Reducing the target size or using a smaller source is the fastest way to speed it up.
Is AI used when shrinking too?
No, and it should not be. Shrinking throws pixels away, which high-quality resampling already does well. Running a model would only cost time without improving anything.
Can I hit a specific file size?
Not here — Resize controls dimensions, not bytes. Use Compress, which lets you set a target such as 200 KB.
How do I avoid distorting the ratio?
Scale and longest-edge modes always keep the ratio. Only exact-pixel mode can distort, and only if you deliberately enter dimensions that do not match.