PDF Toolkit
Merge PDFs, extract or delete pages, rotate and reorder them — processed locally, never uploaded.
Image, PDF and DocumentsImage Optimizer
Resize, crop, compress and convert images in batches — without uploading them anywhere.
Runs in your browser. Nothing is uploaded — your input is processed on this device and discarded when you close the tab.
Optimize images
Drop up to 20 images here, click to browse, or paste an image from the clipboard (Ctrl+V / Cmd+V) anywhere on this page. JPEG, PNG and WebP are accepted, up to 25 MB per file — plus AVIF in browsers that can decode it. Everything is processed on this device only; nothing is uploaded.
No images yet
Output format
Checking what this browser can do with AVIF…
This tool re-encodes every image through the browser canvas, and a canvas re-encode inherently drops EXIF data — camera model, GPS location, timestamps — which is exactly why stripping works here. The photo’s orientation is honoured while it is decoded (via createImageBitmap image orientation where available), so pictures do not come out sideways; they simply carry no metadata home. Un-ticking the box cannot preserve EXIF in this browser-only version, because canvas has no way to write it back — the honest position is that output from this tool never contains the original metadata.
Resize and crop
Crop in this version is centre-based: the largest rectangle with the chosen ratio is cut from the middle of the image. There is no drag-to-reframe yet, so check the before/after preview if your subject sits near an edge.
50% halves both dimensions, which quarters the pixel count. Values that would produce a dimension below 1 pixel are clamped to 1 pixel, and the result line tells you when that has happened.
Compression
Results
No results yet. Add images, choose your settings, then choose Optimize images.
Before / after
Use “Compare before / after” on any result card to load that image here. Judge quality at full size in the downloaded file — previews are scaled by your browser and can hide or invent artefacts.
How to use the Image Optimizer
- Add your images. Drag up to 20 JPEG, PNG or WebP files onto the dropzone, or click it to browse — AVIF files are accepted too in browsers that can decode them. Each file may be up to 25 MB. Anything that fails those checks is refused individually, with the reason written next to its name, and the rest of the batch still loads.
- Choose the output. Keep each file’s format or convert the whole batch to PNG, JPEG or WebP — or to AVIF where this browser can encode it; the option is disabled with an explanation where it cannot. Leave “Strip metadata” ticked unless you have read the note beside it and understand that this browser pipeline removes EXIF either way.
- Resize and crop, if you need to. Resize to exact pixels (width, height, or both, with the aspect-ratio lock on to avoid stretching) or scale by a percentage. Pick a crop ratio to cut the largest centred rectangle of that shape — 1:1 for avatars, 16:9 for video stills, 9:16 for stories.
- Pick a compression style. Either set one quality value for the batch, or switch to compress-to-target and pick 20 KB, 200 KB, 1 MB or your own custom size in KB. Target mode tests real encodes per image, so a detailed photo and a flat logo in the same batch can legitimately end up at different quality settings.
- Optimize and inspect. Choose Optimize images and watch each file move from ready to done. Every result card shows the thumbnail, the dimensions and file size before and after, and the percentage saved — including when the percentage is negative and the output is honestly larger than the input.
- Compare and download. Load any result into the side-by-side before/after view and check faces, text and edges carefully. Download files one at a time from their cards, or take the whole batch at once with Download all as ZIP. Reset clears the batch, and it asks first whenever results exist that you have not downloaded.
Worked example: shrinking the sample photo for a web page
The button below generates a sample image entirely in your browser — a 1200 × 800 gradient scene with shapes and text, drawn on a canvas and loaded into the batch as sample-photo.png. No file is downloaded from anywhere to create it, which makes it a safe way to learn the tool before trusting it with real photographs.
With the sample loaded, try the web-page pass first. Set “Convert to” to WebP, leave the compression on “Use a quality setting”, set quality to 75, and choose Optimize images. Look at three things on the result card: the output dimensions should still read 1200 × 800, because you asked for no resize or crop; the output size should be dramatically smaller than the PNG you started with, because a lossless PNG of a gradient scene is exactly the kind of file WebP was designed to beat; and the before/after preview should show the gradient and circle edges still smooth. We deliberately do not quote an exact output size here: the bytes depend on your browser’s encoder, and any tool that promises you an exact figure in advance is guessing.
Now Reset, load the sample again, and switch compression to Compress to a target file size at 20 KB. Optimize again and read the note on the result card: it tells you the quality the search settled on, and whether the dimensions also had to step down to fit. That note — not the target you typed — is the truth about the file you are downloading, and learning to read it is the whole skill this example is teaching.
What the Image Optimizer does, and why it matters
Images are usually the heaviest thing on a web page, and the cost is paid by everyone who loads it: in data charges on a metered connection, in waiting time on a slow network, and in the storage and bandwidth bills of whoever hosts the page. The awkward part is that the images most people start with — straight from a phone, a camera or a design tool — were never made for that job. They are far larger in pixels than any slot they will occupy, they are saved in formats chosen for editing rather than delivery, and they carry metadata nobody reading the page will ever see. An optimizer is the step that closes that gap: it re-encodes each image at the dimensions, format and quality the use actually requires, and throws away everything else.
This tool does that work in an order that mirrors how the savings actually stack up. Resizing comes first, because pixels are the dominant cost — halving the width and height quarters the pixel count before compression even starts. Cropping to a ratio belongs in the same pass, so a phone screenshot destined for a square avatar slot is not shipped as a tall rectangle and cropped again by every visitor’s browser. Format conversion comes next, because the same picture can differ several-fold in size between PNG, JPEG, WebP and AVIF with no visible difference at sensible quality settings. Compression, whether you drive it by quality or by a target file size, is the final squeeze, applied to an image that is already the right shape. Doing any one of these steps alone helps; doing them together, in one pass, over a whole batch, is what turns a folder of camera originals into a folder a website can afford to serve.
Two design choices are worth naming plainly. The first is that all of this happens in your browser. Your photographs are decoded and re-encoded on your own device, which means the tool works on images that should never be uploaded to a stranger’s server — family pictures, unreleased product shots, scans of documents — and it means there is no queue, no upload progress bar and no copy of your file sitting in someone else’s storage waiting to be deleted. The trade-off is equally real: your device does the work, so very large batches are limited by its memory, and patience is part of the tool on older hardware. The second choice is honesty in the reporting. Every result shows its true before and after sizes, including the embarrassing cases where re-encoding made a file bigger, and target-size mode reports the smallest file it could actually produce rather than pretending every target was met. An optimizer you cannot audit is just another black box; one that shows its working lets you learn, image by image, where your bytes were really going.
Common errors and limitations
- “The output is bigger than the original.” This is real, and the result card shows it as a negative saving rather than hiding it. It usually happens when an already well-compressed WebP or JPEG is re-encoded as PNG, or at a higher quality than the original. Convert to WebP or lower the quality and run it again.
- Crop cut off my subject. Cropping in this version is centre-based — the rectangle is always taken from the middle of the frame. If your subject sits near an edge, crop loosely in your photo editor first, or skip the crop here and keep the full frame. The before/after preview exists precisely so you catch this before downloading.
- AVIF is greyed out or refused. AVIF support is detected, not assumed: some browsers can show AVIF but cannot create it through canvas, and a few cannot open it at all. The note under the format selector states what your browser can and cannot do. WebP is the fallback that keeps almost all of the saving.
- The target size was not met. A noisy, detailed photograph has a floor below which it cannot shrink at a given size — the search lowers quality first, then steps the dimensions down, and then stops and tells you the smallest honest result. Shrinking the image further yourself, or accepting a larger target, are the two real options; no setting can repeal that floor.
- Transparent areas turned white. JPEG and AVIF have no transparency channel in this pipeline, so transparent pixels are flattened onto white when you convert to JPEG. Keep PNG or WebP output for logos, stickers and screenshots that rely on transparency.
- Hard limits. At most 20 images per batch and 25 MB per file. Processing is sequential and happens in your browser tab, so a batch of twenty large photographs is bounded by your device’s memory — if a tab struggles, split the batch. Animated images are treated as a single still frame, and HEIC files from iPhones cannot be decoded by most browsers: convert those to JPEG first.
Frequently asked questions
Are my images uploaded to a server?
No. Decoding, resizing, cropping, compression and ZIP-building all run in this browser tab, on your device. Nothing is sent anywhere, no copy is kept, and closing the tab discards the batch and every result. That is why the tool is safe to use on private photographs and unreleased work — and why very large batches are limited by your device’s memory rather than by a server plan.
Which format should I choose: JPEG, WebP, AVIF or PNG?
For photographs on the web, WebP at quality 75–85 is the safe default: it is widely supported and usually noticeably smaller than JPEG at the same visible quality. AVIF can be smaller again, but only some browsers can create it through canvas and older software may not open it, so use it when you know your audience is current. JPEG remains the compatibility choice for email and older systems. PNG is lossless and much larger — reserve it for images that must stay pixel-perfect, or that need transparency at maximum fidelity, and expect the result card to show little or no saving.
How does compress-to-target decide when to shrink the dimensions?
It searches quality first. For each image it tests real encodes in your browser, looking for the highest quality setting whose file fits the target at full size. Only if even the lowest quality misses does it step the width and height down in fixed stages and search again at the smaller size. If no combination fits, the result card says the target could not be reached and names the smallest file it could honestly produce — the download is that real file, never a padded or mislabelled one.
Does stripping metadata change how my photo looks?
No. Metadata — camera model, exposure settings, GPS location, timestamps — lives alongside the pixels, not inside them, and removing it does not touch the image itself. The one visible risk is orientation: some cameras store a sideways picture plus an instruction to rotate it. This tool honours that instruction while decoding, using the browser’s image-orientation support where available, so the re-encoded output is already the right way up and no longer needs the instruction it has just discarded.
Why did my optimized file come out larger than the original?
Because re-encoding is not automatically shrinking. If the original was already an efficient WebP or a carefully compressed JPEG, saving it as a lossless PNG — or at a higher quality than it started with — produces a bigger file, and the result card reports that as a negative percentage saved rather than disguising it. The fix is to match the output to the job: WebP or JPEG at a moderate quality for photographs, and no format change at all for files that are already small.
Can I reframe a crop instead of taking the centre?
Not in this version, and the page says so rather than implying otherwise: every preset ratio cuts the largest rectangle from the centre of the image. For most product shots, screenshots and centred subjects that is exactly right. When the subject sits off-centre, do the reframing crop in your photo editor first and use this tool for the resize, format and compression pass — or skip the crop here and keep the full frame.
Related tools
Text and Document Comparison
Diff two documents side by side or inline, with change navigation, summary counts and an exportable HTML report.
Image, PDF and DocumentsAdd Page Numbers
Add "Page 2 of 3"-style numbers to a PDF in your browser, with six positions and a custom start count.
Image, PDF and DocumentsRelated articles
How to Optimize Images for Websites
A repeatable image optimisation workflow — size first, then format, then compression, then responsive delivery and lazy loading — with byte sizes measured on real test files, not quoted from other guides.
Choosing Between JPEG, PNG, WebP and AVIF
JPEG, PNG, WebP or AVIF? The honest answer depends on the image type. Includes a size table measured from our own encodes of a photo, a screenshot and a logo — the three types behave completely differently.
How to Resize and Crop Images for Different Layouts
Resizing shrinks a whole image; cropping changes what it shows. Learn when each is right, the aspect-ratio maths with measured crop boxes, a working srcset width set, and exact sizes for common layouts.