Capture session lifecycle

FaceCaptureSession is an EventEmitter with the following shape:


idle → initializing → ready → detecting → hold_steady → capturing → (quality_check) → completed
                                                                                     ↘ retry → detecting
                                                                                     ↘ error

You call start(). The session opens the camera, runs detection, emits a 'guidance' event on every frame, fires 'captured' when a frame passes the gates, then 'completed'. Call stop() to tear down early.

Events

Event	Payload
`started`	none
`state_change`	`FaceCaptureState`
`detection`	`FaceDetectionEvent`
`guidance`	`GuidanceState`
`face_quality`	`FaceQualityResult`
`face_quality_failed`	`FaceQualityResult` (a retry attempt failed)
`face_quality_error`	`{ reason }`
`captured`	face capture result (see note)
`completed`	final result
`error`	`SladeIDError` subclass
`stopped`	none

The captured payload is a partial object — { imageBase64, antispoofScore, deviceInfo } — which you pass straight to enrollFace / matchFace / searchFace; the full FaceCaptureResult (including livenessCompleted, challengeResults) is carried by the completed event.

Reading the state

Wire 'guidance' to the UI for prompts (“hold steady”, “move closer”). Wire 'state_change' for higher-level UI changes (showing a countdown, swapping the overlay). Wire 'captured' and 'completed' for the success path; wire 'error' for the failure path.

Used in

Every capture guide.
Capture sessions reference.