Free Image Tools to Compress, Resize & Convert

Image tools that keep your files on your device

Almost every image utility on this hub runs in your browser using the Canvas API. Your file is read into memory on your own machine, transformed there, and handed back to you as a download. It is never uploaded to us, which is why these tools work on an aeroplane and why there is nothing on our side to leak, sell, or be subpoenaed.

That matters more than it sounds. Most free online converters work by uploading your file to a server, processing it, and serving it back - which means unreleased product photography, client work under NDA, screenshots containing customer data, and personal photos all end up sitting on infrastructure you know nothing about, under a retention policy you did not read. For a lot of professional work that is simply not an acceptable trade for saving thirty seconds.

The trade-off is honest and worth stating: local processing is bounded by your device. A very large batch or a very high-resolution file can be slow on an older phone, and the browser can run out of memory where a server would not. If a tool stalls on a 60-megapixel RAW export, that is the limit you are hitting, not a bug.

Built for the sizes social platforms actually use

Every platform crops. The only question is whether you decide where the crop lands or the platform does. Uploading a landscape photo to a feed that expects portrait means an algorithm picks which third of your image survives, and it is usually not the third with the face or the product label in it.

The canvases worth memorizing are small in number. Instagram feed portrait is 1080×1350 (4:5) and square is 1080×1080. Stories, Reels, TikTok, and YouTube Shorts all share 1080×1920 (9:16), which is why one vertical master can serve all four if you leave margin at the top and bottom. YouTube thumbnails are 1280×720 (16:9). Pinterest favours tall 2:3 pins at 1000×1500. Link previews across Facebook, LinkedIn, and X land near 1200×630.

The resizer and aspect ratio calculator here exist so you export to those numbers before upload rather than letting each app scale and recompress something approximate. Every extra round of platform-side scaling costs a little quality, and it compounds when a file gets cross-posted three times.

Choosing a format: JPG, PNG, WebP, and the rest

JPEG is for photographs. It throws away detail your eye is bad at noticing, which is exactly the right trade for a picture of a face, a meal, or a landscape, and exactly the wrong trade for anything with hard edges.

PNG is for graphics with sharp boundaries - logos, screenshots containing text, diagrams, anything needing transparency. Save a screenshot of a spreadsheet as JPEG and you will see grey fuzz crawling around every letter. That fuzz is the compression doing its job on content it was never designed for.

WebP does both jobs at meaningfully smaller file sizes and is supported everywhere that matters in 2026. It is the sensible default for images you host on your own website. Social platforms will re-encode whatever you give them anyway, so for uploads the format matters less than the dimensions.

HEIC is what modern iPhones shoot by default, and it is the format most likely to be rejected by an upload form or a CMS. Converting to JPEG or WebP before you need it is usually less annoying than discovering the problem mid-deadline.

Compressing without visible damage

Compression quality is not a linear dial. Dropping from 100 to about 80 typically removes more than half the file size while being genuinely hard to see. Dropping from 80 to 60 saves much less and starts showing visible artefacts in gradients - skies, skin, and soft shadows are where it appears first.

Check the result at full size before you accept it, not in a small preview. The specific things to look at are smooth colour transitions, any text in the image, and the edges where a subject meets a background. If those look clean, the compression was free.

One rule that saves grief: never re-compress an already-compressed file repeatedly. Each save discards more information permanently. Keep an original at full quality, and export compressed copies from that original every time rather than compressing the compressed version.

Cropping, safe zones, and small-size legibility

Vertical video and Story frames have UI sitting on top of them - usernames and profile pictures at the top, captions, buttons, and progress bars at the bottom. Anything placed in those bands is either hidden or fighting with an interface element. Keep what matters in the middle 70 to 80 percent of the frame.

Profile pictures are cropped to a circle and displayed tiny. A photo that reads well as a square portrait often loses its subject to the circular mask, and detail that is legible on your monitor vanishes at 40 pixels wide in a comment thread. Crop deliberately and then look at the result at actual display size before committing.

Favicons are the extreme version of the same problem: a full logo with a wordmark is illegible at 16×16. Use a single letter, a mark, or a simplified glyph. The favicon generator here produces the standard set including the 180×180 Apple touch icon, but it cannot make a complex logo readable at that scale - that is a design decision you have to make first.

Metadata: what leaves with your file

Photos carry EXIF data - camera model, exposure settings, the date and time, and on most phones the GPS coordinates of where the shot was taken. Screenshots and exports from design tools can carry software identifiers and sometimes author names.

Major social platforms strip most of this on upload, but that is a courtesy, not a guarantee, and it does not apply to files you email, host yourself, or send to a client. If you are publishing photos taken at your home, your child's school, or a client site, remove the location data before the file leaves your machine. The EXIF remover and metadata viewer on this hub let you see exactly what is attached and then strip it.

A repeatable export workflow

Pick the target ratio before you shoot or design, not after. Produce one master file at full quality and keep it. Resize the master to the platform canvas. Compress and check the result at full size. Strip metadata if the image is going anywhere you do not control. Verify the final dimensions. Upload.

The value of doing it in that order is that the master survives. When a platform changes its preferred ratio - and one of them does, roughly annually - you re-export from the master in ten minutes instead of rebuilding assets from compressed uploads.

For a batch, lock your settings on one image first and confirm the output is right, then run the rest through the identical path. Consistency across a week of posts is more visible to an audience than any individual image being marginally sharper.

What these tools will not do

They will not recover detail that is not there. Upscaling a small screenshot produces a larger blurry screenshot, and no browser tool changes that. They will not remove complex backgrounds with the accuracy of a trained model running on a server - the background colour remover here works on flat, uniform backgrounds and will struggle with hair, glass, or motion blur.

They also will not manage your library. There is no cloud storage, no versioning, and no sync, because nothing is uploaded. Save your output before you close the tab; once it is gone from the browser it is gone.

For everything inside those limits - resize, crop, convert, compress, watermark, strip metadata, generate icons - a local browser tool is faster than any upload-based alternative and leaves no copy of your work behind.