Compress Screenshot PNG JPG

2026-09-13

Why screenshots break naive compression

Screenshots are the hardest image type to compress well, and most people do not expect that. A typical capture is a mix of huge flat areas (a white window, a gray toolbar) and razor-thin 1px text edges. Flat regions compress almost for free, but the crisp edges are high-frequency detail that lossy codecs hate. When you push a JPEG too far, the first visible artefact is ringing: a shimmering halo of color around letters and window borders. On a screenshot, that halo lands exactly where the eye looks first.

PNG or JPG: pick the right container

PNG is lossless and keeps every 1px line perfect, which is why it is the right container for flat UI, menus, and code. JPG throws away data to shrink photos, and a screenshot full of solid color is a photo-like image where JPG performs badly: it smears flat panels and adds noise you can see on a plain background. The trap is that people reach for JPG by reflex because the file is a picture. For UI captures, PNG is usually smaller or close to it once you optimise the palette, and it never degrades.

When converting PNG to JPG is safe

• Safe: a screenshot of a photograph, a gradient-heavy chart, or a video frame where the source is already continuous tone.
• Safe: when you must hit a hard file-size limit and the capture has no small UI text, like a full-screen diagram with large labels.
• Unsafe: any capture with thin fonts, 1px borders, or small icons (invoice lines, IDE text, terminal output). JPG will eat the edges.
• Unsafe: when the file will be re-compressed later (Slack and email re-encode), because each pass adds more ringing.

If you do convert, keep the JPG quality above 90 and never convert a converted file a second time. The safe move is to keep a PNG master and only make a JPG copy when a recipient demands it. image-compressor-saas.shop runs the whole step in your browser, so the master never leaves your machine.

Lossless optimisation beats lossy for UI

For flat UI, the win is palette reduction, not quality loss. A screenshot of a settings page may only use 40 distinct colors; packing them into a 256-color (or smaller PNG-8) palette can cut the file by half with no visible change, because the pixels were already a small set of exact colors. True lossless optimisers also rebuild the PNG compression tables, which shaves more without touching a single pixel. Reach for lossy JPG only when the capture is genuinely photo-like.

The retina 2x trap

Retina and high-DPI displays capture at double resolution, and that quietly doubles file size. A 1440px-wide window on a 2x screen produces a 2880px-wide PNG. If the person viewing it has a 1440px display, you can halve the width to 1440px and lose almost nothing visible, because the screen cannot show the extra pixels anyway. Many people ship the 2x file out of habit. Downscaling first, then optimising, is the single biggest size win for screenshot sharing.

How to judge at 100 percent and hit size targets

Always inspect the result at 100 percent zoom, not shrunk in a chat bubble. Look at three things: small text (does it stay sharp or does it blur), icon edges (do they stay crisp), and 1px borders (do they hold or fade). For size targets, a Slack or chat image should land under 1 MB and ideally under 500 KB; a doc embed under 300 KB keeps the page light; an email attachment under 1 MB avoids getting clipped by providers. The free browser-local tools on image-compressor-saas.shop let you check each of these without uploading the shot.