Compress images. Judge the result.

No. Every image is decoded, resized and re-encoded inside your browser using its own image codecs. There is no upload step and no request that carries a file. You can verify it: open your browser's network panel and compress a batch — no image data leaves the page. This is what makes the tool usable on unreleased product photography, client work under NDA, or scanned documents you would not paste into a hosted service.

Up to 80 files per batch, 25MB per file, and 400MB per batch in total. These are limits of what a browser tab can hold in memory rather than a paid tier, so a desktop will comfortably manage the full batch while a low-memory phone may want fewer. Images are processed a few at a time to keep memory in check, and you can add more files after the first batch finishes.

Not here. Every image is inspected for an alpha channel when it is added, and a transparent image is never converted to a format that cannot carry transparency without an explicit warning first. Smart Optimize keeps transparent images in PNG or WebP, both of which support an alpha channel. The common failure elsewhere — a logo silently flattened onto a white rectangle — is specifically prevented.

Smart Optimize inspects each image and chooses format and quality per file, which is the right default for a mixed batch: photographs become WebP, transparent graphics keep their alpha, and already-small files are left alone. Lossless reduces size without a deliberate quality change, which suits logos, screenshots and anything containing text. Custom exposes quality, format, resize and target-size controls directly for when you know exactly what you want.

You can set a target — under 50KB, 100KB, 200KB, 500KB, or your own number — and the studio runs a binary search over quality to find the highest setting that still fits. It is a search, not a guarantee: a very detailed photograph pushed to a very small target may not get there at any acceptable quality. When that happens you get the closest achievable result, clearly labelled as missing the target, rather than a silent failure.

WebP is the right default for photographs on the web: it is meaningfully smaller than JPEG at the same visual quality and is supported by every current browser. Keep PNG when you need transparency or crisp text and edges. AVIF is smaller still and is offered where your browser can encode it, but it is slower to encode and slightly less universally supported. JPEG remains the safest choice for anything that must open in very old software.

Because compressing it would have made it bigger. Images that are already efficiently compressed — a small WebP, an optimised icon — often grow when re-encoded, so the studio compares the output with the input and keeps the original when the output is larger. That file is reported as skipped with the reason, and the original is included in the download unchanged.

Yes, as a side effect of how browser encoding works. Canvas re-encoding discards every metadata block, including EXIF, so GPS coordinates, camera serial numbers and timestamps in the original are not carried into the output. That is a genuine privacy benefit. It also means ICC colour profiles are lost, which matters for wide-gamut photography — there is no browser API that preserves a colour profile through a canvas encode, so no setting here claims to.

Animated GIF and animated WebP are detected and skipped by default, because browser canvas encoding keeps only the first frame — you would get a still image back and might not notice until later. You can override the skip if a still frame is genuinely what you want, but the warning appears first. For real animated compression you need a dedicated encoder, which a browser does not provide.

Because the encoding is done by your browser's codecs, not by a server-side library. Chrome, Firefox and Safari each ship different JPEG and WebP encoders, and they produce different sizes at the same nominal quality. A server-side tool using MozJPEG or libavif can sometimes squeeze out more, at the cost of uploading your images to it. This tool trades a few percent of ratio for the guarantee that your files never leave your device.

No. Photographs from phones often rely on an EXIF orientation flag rather than storing rotated pixels, and naive canvas pipelines ignore that flag and produce sideways output. Images here are decoded with the orientation applied to the pixels, so the compressed file looks the way the original did — and it no longer depends on a metadata flag that the re-encode would have removed.

Smaller images make pages load faster, and load speed is one input among many into Core Web Vitals. What this tool reports is the actual byte reduction and an explicitly-labelled bandwidth estimate based on your stated monthly pageviews — it does not predict a PageSpeed score or a ranking change, because image weight is only one factor and no tool can honestly promise either.