Image Resize

The resize allocates a canvas at the width and height you asked for and draws the source image into it. The browser's rendering engine handles the interpolation, which gives smooth results when scaling photographs down. With the aspect-ratio lock on, entering a width computes the height from the original proportions.

You can also enter dimensions larger than the original, but enlarging does not create detail that was never captured — edges come out soft. If you need a large version for print, the reliable answer is to go back to a larger source.

The resized image is written back in the same format as the input: feed it a JPEG and you get a JPEG, feed it a PNG and you get a PNG, with the extension matched accordingly. This avoids the classic surprise of "I only resized it and now everything is a heavyweight PNG".

If you want to change format as well, resize first and chain the format converter afterwards.

Resizing and compressing attack file size differently. A 4000×3000 phone photo displayed 800px wide on a blog carries more than twenty times the pixels it will ever show. Lowering the quality in that state still spends bits on pixels nobody sees.

Resize down to the dimensions you actually need first, then compress if it is still too big. For visible quality per byte, that order wins.

Because the image is redrawn through a canvas, EXIF (capture time, camera model, GPS) and ICC profiles do not survive the resize. That is convenient from a privacy standpoint and inconvenient if you wanted to keep the capture data. Files are handled up to 50 MB each.

For something like a social avatar — "square, at this exact size" — resizing straight away distorts the picture. Crop the region you want as a square first, then resize to the required dimensions, and the subject keeps its proportions while the specification is met.