Questions about the thumbnail and metadata viewer

Nothing at all until you press Fetch. At that moment your browser makes anonymous GET requests for public files: the thumbnail images from i.ytimg.com and the oEmbed record from www.youtube.com. The requests carry no cookies and no referrer. There is no server of ours in the path, so nothing we run ever sees the link you pasted.

There is no server of ours to send it to. The page is static files; parsing happens in your browser and the fetches go directly from your browser to Google. That is also why the privacy badge on this product does not claim local-only processing: two real requests do leave your machine, and it would be dishonest to imply otherwise.

The size you are looking at on YouTube is not necessarily one of the published files. maxresdefault only exists if the video was uploaded at 720p or better; when it does not exist, YouTube returns a 120x90 placeholder with a success status. This tool decodes the image and reports that as "Not published" instead of handing you a grey rectangle.

No. A thumbnail is a creative work owned by whoever made it, and YouTube publishes it for previews and link cards. Using one to illustrate a link to that video is ordinary practice; reposting it as your own artwork, or using it to imply an endorsement, is not made lawful by being easy. Whether any specific use is fair dealing or fair use is a question for a lawyer, not a tool.

Unlisted videos publish thumbnails, so they generally work. Private, deleted and age-restricted videos do not serve public thumbnails or an oEmbed record to a signed-out visitor, and this tool will not attempt to sign in or work around that.

SRT, WebVTT, SBV, TTML/DFXP, YouTube’s timedtext XML, its JSON3 form, and plain text copied from the transcript panel. The format is detected for you, and whatever goes in can come out as a transcript, Markdown, CSV, SRT or WebVTT.

No, and no web page can. Measured against the live service: video.google.com/timedtext answers with zero bytes; youtube.com/api/timedtext without signature parameters answers with zero bytes; the signed caption URL embedded in the watch page also answers with zero bytes, because the endpoint now requires a proof-of-origin token issued to a real player session; and the watch page itself sends no cross-origin header, so a page cannot even read the HTML that carries that URL. Captions are gated on purpose, unlike thumbnails, which are plain static files. Getting around it means forging a player session from a server, which this tool will not do. What it can do is read a caption URL you have already obtained — restricted to www.youtube.com, so the tool’s published list of contacted hosts stays accurate.

The quickest route works on any video that has captions and needs no tools: open the video, expand the description, click "Show transcript", then select the transcript text and copy it. Paste that into the Captions panel and this tool turns those "0:15" lines into real cues, lists every one with its timestamp, and lets you download a proper .srt. If the video is yours, YouTube Studio downloads SRT, VTT or SBV directly. If you are comfortable in the network panel, filter for "timedtext" and paste either the response or the URL. All of it is read in the page and never uploaded.

That is how YouTube’s automatic captions are written: each cue repeats the previous line with one more word appended, so the text scrolls smoothly on screen. The Captions panel collapses that rolling window by default, keeping only the completed line. Switch it off if you want the file exactly as written.

No, and that is deliberate. There is no media extraction here, no signature handling and no access-control circumvention. If you need a file you are entitled to from a direct link, the Direct Media Downloader is the separate tool for that.

Because it is the host that does not set an advertising cookie for a visitor who never presses play. It is the default here rather than an option to discover. You can switch it off if you specifically need the standard host.

Something between your browser and Google is refusing the request — commonly a content blocker, a corporate proxy or an offline connection. The request goes directly from your browser, so anything that blocks i.ytimg.com blocks this tool too.