Fast Builds / Computer vision
Computer vision builds: when the camera is the interface
Pose and movement tracking, visual review, camera-based interaction and hand tracking, prototyped fast and tried by real users.
The idea
A system that has to see before it can help
Some ideas only work if the software can see: a movement to follow, a gesture to answer, a hand to track inside a headset, an image to check. A computer vision fast build proves the idea on real people, in the real room, before anyone commits to the full build.
What we build
Computer vision builds for the body, the image, the room and the hand
Each starts as a working prototype in the hands of the people who will use it.
Pose and movement tracking
A camera follows the body: a reach, a squat, a posture held. A build can guide a movement, respond to it, or let a whole body drive a game or a display.
Visual inspection and review
A model looks for what a person would look for, flags what it is unsure about and hands the call back to a reviewer. Changes are checked against a set of known images before release, so a fix does not break something that already worked.
Hand tracking in XR
On a standalone headset the hands become the controller: pinch, point, reach. PTR uses it in VR and mixed reality builds, so a learner handles the task rather than a controller.
Where it fits
Games, characters and rooms that respond
Computer vision often sits inside another build: a game, a character that reacts, or a room the headset has mapped.
How a build runs
From the problem to something people can use
The same fast-build rhythm as every other type, with the camera and the light of the real room.
Start with the decision
Bring the problem, the people who will use it and the decision the build has to support. The first working version goes to them for testing in the space it is for.
Test where tracking gets hard
Low light, a crowded frame, a hand behind an object: the prototype is tried in the conditions it will meet, so the limits are found before the full build.
Privacy agreed before testing
Cameras see people, so each build agrees what is captured, what is kept and where it is processed before testing starts. We process data in the locations and under the terms set out in our governance policy.
Then decide on the full build
Once real people have used it, the decision is made on something real. The full build can follow as a rapid custom build or a custom AI experience. Talk to us or book a time with our team.
Got a camera-based idea to test?
