An arrow on a screenshot is the fastest feedback there is, and every operating system draws one for free. What none of them carry is the reply. The annotated PNG lands in a chat, someone answers three messages later, and by the second round nobody can tell which version of the image the comment belongs to. Here are the free ways to mark up an image, and what to do when the image needs a thread attached to it.
Bottom line
For one arrow on a screenshot, use what took the screenshot: Snipping Tool on Windows, Markup on a Mac or an iPhone. Free, no account, and the marks save into the file. For an image several people have to discuss, a mockup, a chart, a bug screenshot, a proof, the file is the wrong container, because a PNG cannot hold a reply. Upload the image behind a link with an annotation layer instead, so every comment and its answer stay pinned to the pixel they are about.
The free tools that already took the screenshot
Windows: Snipping Tool. Press Win + Shift + S, take the capture, click the notification, and the editor opens with a pen, a highlighter, shapes, a text tool, a ruler and an eraser (Microsoft's guide). Save or copy. On raw speed nothing beats it.
macOS: Markup. After a screenshot, click the thumbnail in the corner before it fades and Markup opens on it directly. For a file already saved, open it in Preview and click the Markup button: pen, shapes, text, highlighter, colour picker.
iPhone and iPad: Markup. Tap the screenshot preview and the same pen and shape tools appear, with the bonus that a full-page web screenshot can be captured and marked up in one pass.
Free, no account, and all three write the marks into the image.
Where a marked-up file runs out
The output of every tool above is a PNG, and a PNG has no idea it is part of a conversation.
Replies detach: your arrow points at a button, their answer is a chat message somewhere below the image. Versions fork: they annotate your annotated screenshot and send it back, so now there are two and one of them is stale. Nothing is pinned, so "the spacing under the header" means re-finding the spot every time someone reads the thread. Two rounds in, the image in the thread and the image somebody is looking at are not the same image.
For one arrow sent once, none of that matters. For a design review, a bug report, a proof or a client sign-off, all of it does.
Annotating an image as a shared page
Open marklayer.app and drop the image on it. PNG, JPEG, GIF, WebP or AVIF, up to 25MB. Or paste the URL of an image already online and skip the upload.
Mark it up: arrows, freehand, boxes and circles, a text label, an area capture with a note, or a comment pinned to an exact point.
Press Share.
They open the link in a browser, with no account and no install, and reply under your comments or add their own. If you are both in it at once, the cursors move.
Each mark records where it sat on the image rather than where it sat on your screen, so it stays on the right pixel when the window changes size or the link is opened on a different display.
Which route fits which job
What you're doing
Use
One arrow on a screenshot, sent once
Snipping Tool or Markup. Fastest, and it saves to the file
An image you need as a flattened file
Snipping Tool, Markup or Preview, same reason
A mockup or proof someone has to sign off on
A link-based annotator, so approval and questions stay on the image
A bug screenshot a developer will ask about
A link, so the follow-up question lands on the pixel
A chart or diagram a group is discussing
A link, so replies thread instead of scattering through chat
The honest limits
No annotated image comes out, because the marks live in the link and not in the file.
No text highlights on an image, since there is no text layer to bind one to. A PDF with real text behaves differently, which is covered in how to annotate a PDF.
The element inspector, measure and multi-select are hidden, because they read a page's HTML and an image has none.
Uploads cap at 25MB, and the file is deleted 90 days after the last time the link was opened.
Annotating is desktop only. Opening a shared link to read works anywhere.
Frequently asked questions
How do I annotate a screenshot for free?
With the tool that took it. On Windows, Snipping Tool opens the capture straight into a pen, highlighter, shapes and a text tool. On a Mac, click the screenshot thumbnail before it fades, or open the file in Preview and use the Markup toolbar. On an iPhone or iPad, tap the screenshot preview and Markup opens on it. All three are free and none asks for an account.
Can I annotate an image online without signing up?
Yes. Drop a PNG, JPEG, GIF, WebP or AVIF onto marklayer.app, up to 25MB, or paste the URL of an image already online, and the drawing, arrow, area and comment tools open over it. No account, no email, nothing to install, and the same is true for whoever opens the link you send.
How do I get comments back on an image instead of just sending one?
Share it as a page rather than as a file. When an image sits behind a link with an annotation layer, anyone who opens it can pin a comment to a specific spot and reply under yours, so the discussion stays attached to the pixel it is about. Sending a PNG in a chat loses that: the reply arrives detached from the image and three messages below it.
Can I annotate a design mockup or a Figma export?
Yes. Export the frame as a PNG or a PDF and open it like any other image. For a mockup that has not shipped, Figma's own commenting is the better fit, because it understands frames and prototype flows and the export does not. The export route is for getting a quick pass from someone who has no Figma account.
Can I download the annotated image afterwards?
Not from MarkLayer. The marks are a layer over the image, held in the share link, and nothing is written back into the file, so there is no annotated copy to save. If you need one flattened PNG to attach somewhere, mark it up in Snipping Tool or Preview instead, since both save into the file.
Does text highlighting work on an image?
No. A highlight binds to selectable text and an image has none, even when there are words visible in it. Drawing, arrows, shapes, area captures and pinned comments all work; text highlighting is the one tool with nothing to attach to. On a PDF that has a real text layer it does work.
Is the image resized when I open it?
It is never enlarged. An image wider than the window is scaled down to fit, and anything smaller is shown at its own size, centred. Blowing up a 200px logo to fill a 1400px frame would hand you a blurred wall to annotate, so the pixels are left alone.
Paste any URL at marklayer.app and annotate the live page in your browser. No install, no account, no sign-up. Recipients of your share links don't need to install anything either.