Upload a clip and this reverse video search checks the web for pages that contain matching or similar frames.
Upload a video
Drag and drop your MP4 or WebM file here
MP4 or WebM · up to 100 MB · up to 10 minutes · processed in your browser
Your full video stays in your browser. We extract a small number of frames locally and send only those frames for web matching.
Reverse video search is a way to search the web using a video file as the query. Instead of typing words, you upload a clip, a few keyframes are sampled from inside your browser, and each frame is matched against publicly indexed web imagery. When a page contains the same or a very similar image, that page comes back as a result.
A reverse video search answers the questions people usually ask about an unfamiliar clip: where did this come from, has it been reposted, and which sites are using this footage right now. Because the matching runs on actual pixels rather than titles, hashtags or metadata, it still works when the video carries no caption or watermark at all.
The full reverse video search tool is right above this section — no account, no installation. Below, you will find what the results mean, how the matching works, and what a reverse video search can and cannot prove, so every result comes with the right expectations.
People use a reverse video search for a handful of recurring reasons. Creators check whether their published clips have been reposted without credit. Journalists and researchers trace where a viral clip has spread. Editors look for the longer or original version of a snippet that arrived with no context. And anyone holding a suspicious clip can check which known sites it actually appears on.
Because every frame is checked independently, one reverse video search covers the whole length of the clip — a page showing only a two-second excerpt can still surface, as long as that excerpt overlaps one of the sampled frames.
A reverse image search starts from one still picture you pick by hand. A reverse video search starts from the moving clip itself: several frames are sampled automatically across the timeline, every frame is matched separately, and the per-frame hits are merged into one ranked list. That difference matters in practice — frames chosen by hand are often credits or transitions, while automatic sampling at 10%, 30%, 50%, 70% and 90% of the timeline skips intros and endings and covers the imagery that actually identifies the video.
A reverse video search works best on footage with distinctive visuals — faces, on-screen text, unusual locations, well-known scenes. The more unique the sampled frames, the fewer false matches and the cleaner the result list. Broadcast clips, music videos and viral moments are usually easy to trace because many pages carry the same imagery.
It struggles when the clip is heavily transformed: mirrored, cropped, speed-changed or covered by overlays. Then some frames still match while others no longer do, so treat a partial reverse video search hit — found on two of five frames — as a real signal rather than noise.
One practical tip: if the first reverse video search returns little, upload a different excerpt of the same video. A scene with more detail gives the matcher more to work with, and a second pass often surfaces pages the first one missed.
Upload your video. Choose an MP4 or WebM file up to 100 MB and 10 minutes. It opens locally in your browser — nothing is sent at this point.
Frames are extracted in your browser. Five compressed frames are sampled at even points through the clip, skipping black frames, intros and credits.
The web is searched for matching imagery. Each frame is matched against the public web index, and the hits are combined into one ranked list of matching pages.
Every result is a public webpage where at least one sampled frame — or a very similar image — appears. Treat each result as a possible source: it may be the original page, a repost, or an article quoting the footage. The tool never labels a result as the verified original.
Each result shows how many of the sampled frames matched. Pages hit on more frames are usually the stronger leads — several independent frame matches are hard to produce by coincidence, while a single-frame hit can be a lookalike scene or generic b-roll.
Web matching cannot prove which page uploaded the video first. A result may be older or newer than the true origin, and indexing delays hide recently posted pages. Use matched pages as evidence and compare publication dates yourself before drawing conclusions.
Your full video stays in your browser — it is never uploaded. The reverse video search extracts a small number of frames locally and sends only those compressed frames for web matching; nothing is written to cloud storage, frame data is discarded once the search completes, and file names are never logged. What you search for stays yours.
Yes — that is exactly what this tool does. Upload an MP4 or WebM clip, and the reverse video search extracts a few keyframes in your browser, matches each frame against the web, and lists the pages containing matching imagery. You do not need an account, and the video itself is never uploaded. Because the query is visual, it also finds pages in other languages that reuse the same footage.
No. The full video file stays in your browser. Only a handful of small, compressed frames are sent for web matching, and even those are discarded as soon as your reverse video search finishes. Nothing is stored on our servers.
Not reliably, and we will not pretend otherwise. A reverse video search returns matching pages — possible sources that may be the original, a repost, or a page quoting the footage. Results are ranked by matched-frame count, which is strong evidence but not proof of who posted first.
Common reasons: the video is new and not yet indexed; the hosting account is private; the clip was re-cut, cropped, mirrored or filtered until the frames no longer match; or the sampled frames are simply too generic — dark scenes and plain backgrounds match nothing distinctive. No results never means the video is unique.
Results can include publicly indexed pages from these platforms whenever matching imagery is available to the web search provider. A reverse video search does not query any platform's private catalog, so unlisted or freshly uploaded videos generally cannot appear.
Yes, the tool is currently free and needs no sign-up. Each search processes a small number of frames, which keeps the reverse video search fast and focused on strong visual matches.
The same engine, tuned for a different question.
Search by video: upload a clip and the web is searched using frames extracted from it.
Upload a clip to find video source candidates: the webpages where its matching frames appear.
Upload the clip and check where its frames appear — the answer to what video is this from is usually one click away.