Broadcaster
A broadcaster with a large archive or high-volume live production who cannot staff an editor for every social cut-down
AI & automationReady to pilotBuilt by Labs and Cloud Video KitLast updated 29 July 2026
Video VerticalizationAutomatic reframing of horizontal footage into vertical video
Social cut-downs wait in an editor queue, so the clip publishes after the moment has passed. Video Verticalization reframes horizontal footage into a clean 9:16 automatically, tracking active speakers, faces, people and salient objects frame by frame so that social and mobile versions ship alongside the original without a manual step.
Automatic Subject CenteringPlatform-ready 9:16Vision-based Intelligence
Watch the demo
Original (16:9)
Click to unmute
Verticalized (9:16)
The problem this solves
Broadcasting a match or event is one workflow; distributing the same content across social platforms is a second, separate production job. An editor has to select the clip, open it in a timeline, make crop decisions for every scene change, and export. By the time that is done, the window when engagement is highest has closed.
The bottleneck is not talent or tools - it is that reframing is treated as an editorial decision rather than a mechanical one. For most clips the right framing is obvious: keep the active speaker or the moving subject in frame. A system that can make that call from the video itself removes the queue entirely.
Who this is for
This is relevant if you are:
Sports organisation
A sports organisation or rights holder who needs vertical clips of every fixture moment to land on platforms within minutes, not hours
Buyer
A publisher or content owner converting a back catalogue for vertical-first distribution
How it works
- 1Each frame is analysed to identify the foreground subjects - active speakers, faces, people - and to suppress the background elements that should not drive the crop decision.
- 2Dedicated face and person trackers follow those subjects across consecutive frames, so one subject stays one subject through movement, occlusion and re-entry rather than being rediscovered every frame.
- 3Scenes with no people fall to a salient object detector, which keeps the crop on the visually dominant element instead of defaulting to the centre of the frame.
- 4The reframing engine adjusts the crop window continuously to produce a 9:16 output, and handles the cases that trip up simpler approaches: at a scene change or a speaker transition it shifts between subjects without cutting, so the clip keeps its context.
- 5The output is a vertical video file ready to post, with no manual keyframing or editorial decision at any point in the pipeline.
What makes it different
Rule-based centre-crop produces vertical video that looks wrong whenever the subject is not in the middle of the frame, which is most of the time in live production. This system makes a framing decision from the content of each frame rather than from a fixed crop position. It also handles the transitions that trip up simpler approaches: speaker changes, scene cuts and crowd shots with no clear primary subject all get their own logic rather than falling through to a default.
Current status & next steps
Ready to pilotReviewed 29 July 2026
Video Verticalization is ready to pilot. The demo on this page runs against a set of prepared clips and shows the full reframing pipeline operating on real footage.
The pipeline performs best on scenes with a single primary subject or a small number of people. High-density scenes with many individuals reduce framing accuracy and are not yet fully optimised.
The next step is a pilot against a customer's production content, so framing accuracy can be measured on the source material and shot types that actually matter to them.
Want vertical versions of your content without the queue?
Tell us roughly what you produce and which platforms you are targeting, and we will tell you what a pilot against your material would involve.