Compress Screenshot — Free, Fast & Private
A screenshot is nothing like a photo: hard-edged text, single-colour panels, one-pixel borders, and not a gradient in sight. Run it through a photo compressor at JPEG quality 60 and the letters grow halos, the flat panels sprout blotches, and the file is small but the screenshot is useless — you can no longer read the error message you captured it for. The right approach compresses the parts that are cheap to compress and leaves every glyph exactly where it was.
The files are deceptively large, which is why people reach for a compressor in the first place. A single full-screen capture on a modern display can be 3-8MB as PNG, and Retina/HiDPI screens double or triple that. Attach a few of those to a Jira ticket, a Slack thread, a Confluence page or an email and the size adds up fast. Captured on a phone instead? You can compress images on an Android phone without an app.
For most screenshots, PNG with Smart compression is the best approach. It preserves every pixel of text while removing unnecessary metadata and optimising the colour palette. This typically reduces file size by 40-60% without any visible change. If you need even smaller files, converting to JPEG at quality 85+ works well for screenshots without transparency — the text stays readable at that quality level.
Retina screenshots deserve special attention. A 2x Retina capture of a 1440px-wide screen produces a 2880px-wide image. If the screenshot is going into documentation or a web page that displays at 1440px, you can resize to 1920px max width and still have more than enough detail. This single step cuts file size by over 50% before compression even starts.
All processing runs in your browser. Screenshots often contain sensitive information — code, internal tools, private conversations, customer data. With MiniPx, none of that leaves your device. Sometimes the right compression is total: when what you actually need from a capture is its words — an error message, a snippet from a slide — extract the text with on-device OCR and paste kilobytes of text instead of megabytes of pixels. If a screenshot has to go public with a detail that should not — an account number, an email address, a customer name — pixelate that region before compressing, and use generous block sizes: heavily pixelated text is the only kind that stays hidden. If the detail is a face — a video-call tile, a colleague's webcam corner — the face blur tool can find and soften every face on-device in one pass. And when a light-mode capture needs to sit in dark-mode documentation, a quick pass through the colour inverter makes a serviceable first draft.
If a screenshot contains red or blue text and looks fringed after saving as JPEG, the cause is chroma subsampling rather than the quality setting — JPEG stores colour at half resolution, so PNG or WebP is the right format for coloured type.
For screenshots that live in documentation rather than a chat, the documentation-specific guide covers retina capture and readable UI text.
A PNG screenshot with a hard cap is covered on compress PNG to 100KB.