Obsidian Web Clipper: how to save images locally

Why Web Clipper does not download images

The omission is deliberate. Web Clipper hands the note to the Obsidian app, and text travels well that way. Image files do not.

Users asked for it early. The GitHub issue "Download pictures to local", opened in September 2024, collected 32 thumbs-up before being closed. The answer from Obsidian's side came in two parts: embedding images as base64 was ruled out because it bloats notes to several megabytes, and the download itself was moved into the Obsidian app. In April 2026, a pull request adding image download directly to the extension was closed with the comment that the idea had "already been evaluated and rejected in the past".

So the official path runs through Obsidian itself.

Method 1: Obsidian's built-in command

Obsidian 1.8.0, released on December 18, 2024, added a command for this. From the release notes: it "downloads all externally embedded images" and "replaces the external links with internal embeds".

1. Clip the page with Web Clipper as usual.
2. Open the new note in Obsidian.
3. Open the command palette: Ctrl+P, or Cmd+P on a Mac.
4. Run Download attachments for current file.

The images land in your vault's attachment location, the one set in Settings, Files and links, and the note now embeds the local copies. If you clip often, assign the command a hotkey in Settings, Hotkeys.

This works well for public pages: blog posts, documentation, news articles.

Where it breaks

The command runs inside Obsidian. It requests each image as a stranger would, without the cookies of your browser session and without the referrer of the page you clipped. Two kinds of images suffer:

• Images behind a login. Paywalled articles, private wikis, internal tools. One user on the issue tracker describes ending up with "a login page saved as a .png".
• Images protected against hotlinking. Some CDNs refuse requests that do not come from the original page.

It is also a manual step, once per note. Forget it, and the note silently keeps its remote links.

Method 2: a community plugin that does it automatically

Community plugins can run the download for you. Local Images, which Obsidian's own team pointed to before the built-in command existed, finds external image links in a note, downloads the files into the vault and rewrites the links. Other plugins watch your clippings folder and process each new note as it arrives.

That removes the manual step. It does not remove the blind spot: these plugins also fetch from inside Obsidian, so images that need your browser session fail the same way. Check a plugin's last update before depending on it.

Method 3: download the images at clip time, from the browser

The only moment when every image is reachable is while you are looking at the page, signed in, in your browser. A clipper that downloads images at that moment sidesteps the session problem.

Markforge is a Chrome extension that works this way. It is a different clipper, not an add-on to Web Clipper, so this route means clipping the page with Markforge.

1. Install Markforge and open its settings with the gear icon.
2. Under Download Folder, choose a folder inside your vault.
3. Under Compatibility, turn on Use Obsidian wikilink syntax for images.
4. On the page you want, press Alt+M, then click Download .md.

The note is written to your vault folder, and its images go into an images subfolder next to it, embedded as ![[images/file.png]].

For each image, Markforge tries three things in order:

1. Fetch it from the extension and save it as a file.
2. If that fails, fetch it from inside the page, the way the page itself loads it, session included. An image retrieved this way is embedded in the note as base64, up to 800 KB.
3. If both fail, keep the original link, so nothing is lost silently.

The popup reports the outcome when it is done: how many images were saved as files, how many were inlined, how many were skipped.

Two filters keep the folder clean. Images smaller than 100 pixels are ignored by default, which removes icons and avatars, and files over 5 MB are left as links. Both thresholds are adjustable in the settings.

What you give up

Markforge is not Web Clipper. It has no highlighter, no reader mode, no template logic, no AI interpreter, and it runs in Chrome only. It writes a Markdown file and its images, and that is the extent of it. The base64 fallback also carries the cost Obsidian's team warns about: a note with several inlined images gets heavy. It is a fallback for the stubborn cases, not the default.

Which method to use

Many people will end up combining the first and the third: Web Clipper for everyday reading, and a browser-side download for the pages where images matter and refuse to come along. For a wider look at the options, see our comparison of Markdown web clippers.