No GPS needed.
No metadata needed.
Upload. Extract. Locate.
Where was this
video taken?
Drop the video in. Oceanir pulls the clearest frames out of it for you, then reads the buildings, roads, signage, and terrain to estimate where it was filmed. No GPS. No metadata. Just the pixels.
Why you need to know
The video exists. The question is where. The metadata is gone (it usually is). The account is anonymous. What you have is the footage itself, and that is enough.
A clip surfaces on social media claiming to show an event in a specific city. The account is anonymous. The video metadata is gone. You pause on the widest outdoor frame and need to confirm where it was actually filmed.
A source sends you a video of something they say happened abroad. Before you publish, you need to verify the location independently. You have no GPS, no timestamp, just the frames.
A piece of archival footage shows a street scene with no location record. The buildings, the road markings, the shopfronts. You need to identify the city and country before you can use it.
You watched a travel video and paused on a shot of a beautiful old city street. The creator did not say where it was. You want to know.
How to find where a video was filmed
Three steps. The middle one is where the real work happens.
Hand over the video
Drop the clip straight into Oceanir. It pulls candidate frames out of the video for you, and you pick the one that shows the environment: the street, the buildings, the background. Not a close-up. Not a dark scene. Not a blurry pan. You want the moment the camera pulls back and the world becomes visible.
If you would rather pick the frame yourself, that still works: pause in any player, screenshot, and upload it as a JPG, PNG, or WebP.
Upload to Oceanir
Drag and drop the frame into Oceanir or click to upload. No account needed to start. The AI reads the frame the same way it reads any photograph: looking at road markings, building styles, signage languages, vehicle types, vegetation species, and shadow angles to narrow the world down step by step.
Important: Oceanir does not read GPS coordinates, EXIF tags, or any metadata from the frame file. The analysis is based entirely on what is visible in the pixels. This means it works on screenshots, re-saved frames, and social-media grabs where metadata was stripped long ago.
Review ranked candidates
Oceanir returns ranked location candidates, not a single pin. Each candidate comes with a confidence score, the reasoning chain that supports it, and a Street View comparison so you can compare the frame against the proposed location visually. You make the final call. The model narrows the world. You close the case.
You can also cross-reference results with other image geolocation workflows by uploading different frames from the same clip.
Choosing the right frame
The frame you pick determines the result. One good frame beats ten mediocre ones. Here is what to look for.
Wide establishing shot
The moment the camera pulls back to show the full street or skyline. More geography visible per pixel.
Storefront or signage
Text on signs, shop names, street names. Even a partial word can narrow a country or city.
Intersection view
Road markings, traffic signals, pedestrian crossing styles, and lane widths vary by country.
Background in focus
Shallow depth-of-field shots blur the background. Choose a frame where the environment behind the subject is sharp.
Daytime and well-lit
Night frames hide architecture and signage. A well-lit daytime frame contains far more geographic signal.
Tip for news and social clips: Social-media video is often compressed and cropped. Look for a moment when the creator accidentally shows a street name, a shop sign, a bus number, or a recognizable building in the background. Those incidental details are often the strongest geographic signals.
What the AI reads in one frame
A single frame can contain dozens of geographic signals. The model reads them simultaneously, narrowing the search until a confident location emerges. Here is what that process looks like.
Reading frame...
Waiting...
Waiting...
Waiting...
Waiting...
Waiting...
Waiting...
Which videos work best
Strong candidates
- +Travel vlogs with outdoor establishing shots
- +News footage showing streets and buildings
- +Social clips where the background is in focus
- +Dashcam footage of city driving
- +Documentary footage with wide exterior shots
- +Security camera exports showing outdoor scenes
Harder to locate
- –Indoor footage with no exterior windows
- –Nighttime video where details are dark
- –Highly compressed clips where text is unreadable
- –Talking-head clips with blurred backgrounds
- –Animation or screen-recorded content
- –Aerial footage from very high altitude
Find where it was filmed
Upload the video itself — frame extraction happens here, not in your video player. Get ranked location candidates with confidence scores and Street View verification. Free to try, no credit card required.
Start analyzingCommon questions
Yes. Upload the video file itself and Oceanir extracts candidate frames from it for you, then reads the visual content of the clearest one to estimate where it was filmed. You can still upload a single still frame as a JPG, PNG, or WebP if you would rather choose it yourself. No GPS or metadata is needed either way.
Choose a wide outdoor shot that shows as much of the environment as possible. Good frames include visible streets, building facades, storefronts, signage, terrain, or skyline. Avoid tight close-ups, blurry transitions, night shots with limited detail, and frames that are mostly sky or interior. The more geographic information is visible, the better the result.
Yes. Short clips (TikTok, Instagram Reels, YouTube Shorts, X/Twitter video) often contain clear outdoor frames. Upload the clip and Oceanir pulls frames out of it, then you pick the one with the most visible geography. Compressed social video is fine as long as a clear frame exists.
Upload the video and Oceanir extracts the frames server-side, so MP4, MOV, and the common web formats work without you touching a video player. Very unusual containers or codecs can fail extraction; if that happens, take a still frame yourself and upload it as a JPG, PNG, or WebP.
Accuracy depends on the visual content of the frame. Outdoor scenes with distinctive architecture, road markings, signage, or terrain produce the best results. On the Im2GPS3k benchmark, Orca 1.5 achieves 32.2% accuracy within 1 km and 64.3% within 25 km. Oceanir returns ranked candidates with confidence scores so you can assess which result is most supported by the evidence.
No. Oceanir performs pure visual analysis on the frame. It does not read EXIF data, GPS tracks, camera metadata, or any embedded information. This is by design: most social-media platforms strip metadata on upload, and many video files never had GPS embedded in the first place. The analysis works from the pixels alone.
Free to try, no credit card required. Each analysis uses one credit. Pro includes capped API access for individual development and low-volume internal use. Organizational, client-facing, batch, and production integrations require Enterprise. Paid plans include Starter for occasional D3, Pro for individual verification workflows, and sales-led Enterprise plans.