Oneira Sight is designed around a clear boundary: the owner chooses when to capture one frame. The camera is not presented as a background observer. That deliberate start helps people understand when visual data leaves the device and when no Sight request is happening.
One frame, one visible purpose
The current Sight input is one deliberate frame attached to a focused owner-selected purpose. That purpose narrows the requested response before the image is processed. The output is a short on-screen reply, not a continuous video analysis, hidden archive, or open-ended surveillance feed.
Consent comes before the button
The owner must use Sight only when everyone present has agreed to the camera use. A physical capture action proves that the owner acted; it does not prove that another person consented. The human conversation remains necessary, and a refusal should end the capture attempt without pressure.
Cloud processing remains part of the flow
The frame uses an optional connected service for the reply. JML states that it does not store the image or reply in this release, while connected processing follows the active service's data controls. Important results should be checked because connected visual help can still be incomplete or wrong.
The short version
- Sight begins only after an owner-triggered capture.
- Everyone present must agree before an owner-triggered Sight capture.
- Connected processing applies even though JML does not store the frame or reply in this release.