Product
How WhatMovieIsThis works
Everything the tool does between your upload and your answer, described plainly enough to decide whether to trust it.
Step 1 — You give the tool a signal
There are three input modes. Screenshot mode accepts up to ten images at once, which is the single biggest accuracy lever available to you: frames from different scenes constrain the answer far more than several frames from the same second.
Video-link mode accepts a public video URL, pulls representative frames and, where speech is present, transcribes the dialogue so the visual and spoken signals can be cross-checked. Quote mode accepts a remembered line, a plot description, a character sketch or an actor description as free text.
Step 2 — The model reasons about the content
The frames or text are sent to a multimodal model with a structured prompt that asks it to name the title, state its confidence, and justify the answer from the specific visual or textual evidence available.
This is inference, not lookup. There is no database of every frame of every film — which is exactly why the tool can identify a still that has never been published online, and also why it can be confidently wrong about a lookalike.
Step 3 — The answer is enriched
Once a title is proposed, the result is filled out with release year, genre, runtime, language, director and principal cast, aggregate ratings, a spoiler-light synopsis, similar titles, awards, an official trailer link and legal streaming, rental and purchase options.
Poster and backdrop imagery is resolved from public sources; when no artwork can be found the result renders without it rather than showing a placeholder that implies a match we don't have.
Step 4 — You see the reasoning, not just the verdict
Every result shows a confidence band and a short 'why this title' explanation. If the explanation cites a detail you know is wrong, treat the answer as a hypothesis and run again with another frame — that feedback loop is the intended way to use the tool.
Search history and favourites are stored locally in your own browser so you can return to a previous answer without an account.
Frequently asked questions
Do I need an account?
+
No. Every identification tool on the site works without signing in, and there is no usage quota.
How long does an identification take?
+
Screenshot and text identifications typically return in a few seconds. Video links take longer because frames must be extracted and audio transcribed first.
Can I upload more than one image?
+
Yes, up to ten per identification. Use frames from different scenes for the best result.