counterfactual lens
A vision model predicts a hidden part of the camera scene, while the experience renders the error between prediction and reality as the primary material.
surface
A masked camera region is temporarily replaced by a plausible machine prediction.
disturbance
The visitor moves the mask away and reveals the real scene while the system preserves only the changing difference field.
recognition
Prediction can look continuous with reality until reality is allowed to contradict it. The useful artifact is the structure of the miss.
system
Explicit camera permission, local or privacy-bounded vision inference, image completion only when justified, and procedural visualization of prediction error rather than a chatbot explanation.