Set an exact pixel size, scale by percentage, or press 2×, 3× or 4× to enlarge a picture in one step. The aspect ratio stays locked while you type, so the output keeps the proportions of the source instead of stretching.
A high-quality smoothing mode is used for photographs, and a hard-edged nearest-neighbour mode is available for icons and pixel art. Everything renders on a canvas in your browser, so the file you pick is never uploaded. All processing runs locally in your browser. Your data is not uploaded to any server. 100% Local
Pixel mode is for places that publish exact requirements: a 1200 by 630 link preview, a 1080 by 1080 social post or a 3000 pixel print master. Percent mode is quicker when you only want a lighter copy, because 50 percent halves both edges and cuts the pixel count to a quarter. Both modes still round to whole pixels, since a canvas cannot store half a column.
Interpolation fills new pixels from existing neighbours. It cannot invent detail that the camera never recorded, so a 4× export looks smoother but not sharper when zoomed in. This tool is a classic resampler, not an AI super-resolution model: there is no learned texture synthesis and no face reconstruction. Upscaling is still useful for meeting a minimum pixel requirement, matching a canvas size or making an image easier to annotate.
Each platform asks for a specific frame: 1200 by 630 for a shared link, 1080 by 1080 for a square post, 1080 by 1920 for a story. Resizing to those numbers before uploading avoids automatic crops that cut off text. Enlarge only when the source is slightly under the requirement, because a stretched small image reads as soft on a phone screen.
A print shop usually states both a physical size and a resolution: a 10 by 15 centimetre photo at 300 DPI needs roughly 1181 by 1772 pixels. Use pixel mode, compare the numbers with the source, and enlarge only if the gap is small. A 2× upscale can rescue a slightly undersized file, but it will not turn a phone thumbnail into a poster.
Scaling a 32 by 32 sprite with smooth interpolation turns crisp edges into grey fuzz. Nearest-neighbour repeats whole pixels instead, so the art keeps its blocky character at 2× or 4×. The same setting is handy for screenshots of dialogs and QR codes, where a softened edge can make a thin line or a module unreadable.
No. Scaling up interpolates existing pixels and invents nothing new. A 2x export has four times the pixels but the same real detail, so it looks softer at 100 percent. This page uses ordinary interpolation, not AI super-resolution, and cannot recover information that was never captured.
Every new pixel is guessed from its neighbours, and at 3x or 4x the guesses dominate the frame. High-quality smoothing avoids jagged edges but cannot create texture. If the photo will be viewed large, start from the largest original available rather than enlarging a small copy.
It tells the browser canvas to spend more effort choosing intermediate pixel values instead of a fast approximation. Edges look less jagged when shrinking and less blocky when enlarging. Results still depend on the source pixels, and some browsers implement the hint differently.
Use it for pixel art, icon sprites, QR codes and screenshots of user interfaces where hard edges carry meaning. Smoothing would blur a one-pixel line into grey mush. For photographs, product shots and text in a layout, the smooth high-quality mode is the better default.
Yes. With the ratio lock enabled, editing width recalculates height from the original proportions and rounds to the nearest whole pixel, so the result never drifts. Turn the lock off when a fixed frame such as 1200 by 630 is required. Check important results yourself before publishing.