An extended Reality (XR) system provides methodologies for capturing hand poses being made by a user in low-light environments. The XR system capture, using one or more visible light cameras, tracking video frame data of a hand pose of a user of the XR system. The XR system generates hand-tracking data based on the tracking video frame data. The hand-tracking data includes a skeletal model and a hand-tracking confidence level indicating a probability that the skeletal model matches the hand pose. The XR system compares the hand-tracking confidence level to a threshold confidence value, and based on determining the first hand-tracking confidence level is below the threshold confidence value, activates one or more wide-spectrum cameras to capture subsequent tracking video frame data of the hand pose. The XR system may also activate an IR light emitter to illuminate the hands of the user.