01
Compression
Compression changes how pixel information is encoded while keeping the same width and height. For JPG or lossy WebP, lowering quality usually reduces size, but aggressive settings can create block patterns, banding, or fuzzy detail. Start near 80, inspect the result at normal viewing size, and lower it in small steps rather than jumping to an extreme value.
02
Resizing
Resizing changes the number of pixels and often delivers the largest saving when the original is much bigger than its destination. A 4000-pixel camera photo does not need to remain 4000 pixels wide when a profile image is displayed at 800 pixels. Keep the aspect ratio locked unless the destination requires exact dimensions.
03
Changing the format
Format choice should follow the content. JPG or WebP often works well for photographs; PNG remains useful for screenshots, text, diagrams, and transparency. Converting a photo from PNG to JPG can save a lot, while converting a small flat-color graphic may offer little benefit and can damage sharp edges.
04
A reliable order of operations
First crop away content you do not need. Second resize to the largest dimensions the destination actually uses. Third choose the appropriate format. Finally adjust quality until the file meets the limit. This order avoids spending bits on pixels that will later be discarded.
05
How to judge the result
Compare the processed image at its intended display size, not only when zoomed to 400%. Check faces, small text, product edges, skies, and smooth gradients. Keep the original file separately because optimization is a delivery step, not a replacement for your master copy.