What can you use it for?
Online Image Optimizer is a set of small tools that share one drop zone. Make a photo small enough to email, reshape it for wherever you are posting, turn a PDF into images, strip out where it was taken, darken it so text can be read on top, sign it with a watermark, or convert between formats. Every one of them is below, with what it is for. It is free, there is nothing to install, and there is no account.
The photo is too big to email or text
A photo straight off a phone is commonly 3 to 8 MB. Mail servers bounce it, group chats compress it into mush, and the person receiving it on mobile data will not thank you.
Drop it in and take the small JPEG. In our tests a 6.5 MB holiday photo came back at 17 KB and still looked right on a phone screen — small enough that nothing will refuse it.
You have a whole folder of them
Doing forty holiday photos one at a time is nobody's afternoon. Select up to twenty, drag them over together, and they all get the same treatment in one pass.
Each one keeps its own filename with the width added, so you can match them back up afterwards, and the lot comes down as a single zip. Or combine them into one picture — a collage, a grid, or a contact sheet.
You photographed the pages of a document
A contract, a form, a set of receipts. Eight pages off a phone is around thirty megabytes, and the portal you need to send them to will not take that.
The quality is held higher for this than it would be for a holiday photo, because a face survives heavy compression and a reference number does not. Eight pages typically come back at about a megabyte, still readable.
What you have is a PDF, and what you need is a picture
Plenty of things take an image and refuse a document: marketplace listings, older upload forms, messaging apps, the image field on a web page.
Drop the PDF in and every page comes back as an optimized JPEG, in one zip, in page order. Up to fifty pages at a time.
It needs to be the right shape for Instagram, Facebook or LinkedIn
Every platform crops differently, and it always crops the bit you wanted to keep. Pick a preset instead and decide the framing yourself, or draw the rectangle by hand and set the width in pixels afterwards. For a profile grid, cut one picture into tiles instead.
Square 1:1 and portrait 4:5 for Instagram, 9:16 for stories and reels, 1200 × 630 for a Facebook or Open Graph link preview, 1584 × 396 for a LinkedIn banner, and plain 16:9.
You are putting it on a website
Tick Build <picture> elements and one upload gives you the whole responsive set: JPEG, WebP and AVIF, each at a desktop and a phone size, plus JPEG XL and PNG if you want them. Everybody else just gets one sensible file, which is the point of it being a box you tick.
And a <picture> tag written out for you, with the sources in the right order and the srcset widths filled in. Copy it, paste it, done — every browser takes the smallest file it can read.
You would rather not share where the photo was taken
Photographs carry hidden metadata: GPS coordinates, the date, the camera and often its serial number. Post one straight from your phone and all of that travels with it.
Every image that comes out of here has that stripped by default. You do not have to ask, and you can switch it back on if you actually want it kept.
Your selfie came out the wrong way round
A front camera shows you one thing and saves the mirror of it. The writing on your shirt reads backwards, your parting is on the wrong side, and the face looking back is not the one you are used to — because it is the one everybody else sees.
Drop it in and it is mirrored left to right, back the way you actually look. Top to bottom is there too, along with quarter turns for a photo that came out sideways, and the photo keeps its own size.
Text you can actually read on a photo
White text can be hard to read over a bright picture, and a heavier font weight does not solve the lack of contrast. Hero images, title slides and social cards commonly add a translucent overlay to the photograph so the text remains clear.
Drop in an image and adjust the slider until your headline is clear, then download the result. Black is the default because darkening is the usual job. A white overlay lightens a picture behind dark text, while applying the same brand colour to a set of photographs gives a presentation a consistent look.
You want your name on it
A watermark says who made a picture without hiding it. Photographers, illustrators and anyone posting work they would rather not see reused unattributed put a name, a studio or a web address in a corner of it.
Write the words and choose where they sit, how large they are and how strong, in one of eight typefaces. A logo works too if it is already online: give its address and that is drawn on the picture instead of the words.
You want to put a caption on it
A picture with words across it travels further than the same picture on its own. Group chats, slides, a poster for the noticeboard: the caption is what makes it worth sending, and it has to be readable at thumbnail size before it is anything else.
Type a line for the top and a line for the bottom and they are drawn in heavy capitals with a hard keyline, wrapped and shrunk until they fit whatever shape the picture is. Bring your own photograph, or start from one of the blank cards this page can draw for you.
Something will only accept a different file type
An upload form wants a JPEG and your phone produced something else. A print shop wants a PNG. A colleague sends you a format nothing will open — most often an iPhone photo saved as HEIC, which Windows and most upload forms still refuse. A photograph saved as a PNG can be made a quarter of the size as a JPEG, and the reverse conversion is there for software that insists on it.
Drop it in and take whichever one you need — JPEG, PNG, WebP, AVIF, JPEG XL or GIF — at whatever size suits. No installing anything to convert one file. An animated GIF is the one case where the format has to stay put, and it can still be made smaller with every frame kept. A short video clip can become one too, converted on your own device rather than uploaded.
Going the other way is just as common: a picture saved from a website arrives as a WebP, and the program you want to put it into refuses it. And when the destination takes no picture at all — a README, a commit message, a chat window that strips attachments — a photograph can be rebuilt out of characters instead. A logo that arrives as a vector drawing has the same problem in reverse, and can be turned into a PNG at any size without leaving your browser.
What it is not
It is not somewhere to store images. Everything you upload is deleted automatically after a short while, so download what you want before you close the tab.
It is also not a photo editor. You can select an area to crop, flip an image left to right or top to bottom, and rotate it in ninety-degree steps — the things people actually need before sending a picture. There are no filters, retouching tools, drawing tools or layers, and none of those features is planned.
One image in, everything you need out.
Curious which format actually wins? We measured them: JPEG, WebP, AVIF and JPEG XL compared.