MeshTigator

Based on my own understanding and observations, MeshTigator appears to reconstruct physical space using the LiDAR sensor built into the device, employing a Direct Time-of-Flight (DToF) method. Simply put, it calculates distance by measuring the time it takes for infrared light to be emitted, reflected, and return to the sensor. As a result of technical verification conducted using AI, it is estimated that the iPhone’s LiDAR sensor physically emits approximately 576 infrared points. However, through interpolation processing performed by the Neural Engine, the space is reconstructed with an apparent point cloud density equivalent to roughly 49,000 points per frame. Here, the term “one frame” means that, for example, in a 30 FPS video, 30 frames are displayed per second. However, the internal frame rate at which MeshTigator updates its spatial data has not been publicly disclosed. Therefore, the expression “approximately 49,000 points per frame” does not refer to the internal processing frame rate, but should instead be understood as a way to describe the apparent density of the reconstructed point cloud. After distance data is acquired, the environment is reconstructed as a mesh-based 3D space. This data is then analyzed by the Neural Engine at an extremely large computational scale, and ARKit’s Human Pose Detection is applied only when the structure is interpreted as a human skeletal form. Unlike conventional SLS cameras, MeshTigator does not appear to be directly “searching for humans.” Instead, it detects physical anomalies or objects that exist within the reconstructed mesh space. Non-human-shaped anomalies can also be detected, but stick-figure skeletons are displayed only when the structure is judged to be extremely close to a human posture. Human Pose Detection is designed to eliminate unnatural joint movements, which makes the kind of false-positive stick figures commonly seen with SLS cameras feel extremely rare in this case. LiDAR sensors are widely used in fields directly related to human safety, such as autonomous driving. Among sensors available to general consumers, LiDAR is currently one of the most precise methods for detecting physical space. In addition to Mesh mode, MeshTigator also includes a LiDAR mode. In LiDAR mode, physical changes within the space are visualized in a display format similar to a thermal camera, making subtle spatial differences easier to identify. Multiple display modes exist—CAM, Mesh, and LiDAR—but the LiDAR sensor itself appears to be active at all times. Phenomena often described in paranormal investigations as “unidentified shadows” may appear as subtle changes in distance or reflection characteristics. A shadow itself is not a physical object, and LiDAR does not directly detect shadows. What LiDAR measures are minute differences in infrared reflection caused by factors such as fog, smoke, air density, or temperature variations. As a result, areas that appear to the human eye as simple shadows may manifest as spatial anomalies in LiDAR data. In other words, LiDAR is not capturing the shadow itself, but the physical conditions that produce the shadow. MeshTigator feels far more reliable than traditional SLS cameras, but false detections are not completely eliminated. In rare cases, stick figures may appear outside the effective range of the LiDAR sensor; this behavior is expected, as these areas lack fully constructed 3D mapping. Very rarely, stick figures may also appear on trees or buildings. However, in such cases, approaching the object allows it to be correctly recognized as non-human, and the stick figure disappears. This is an excellent application. Even just launching the app and watching the mesh being constructed is genuinely exciting. I purchased the latest iPad Pro M5 specifically for this app, and I truly believe it was worth it. I am confident that this application will become an essential tool in my future investigations. Finally, on a personal note, I strongly hope that Japanese language support will someday be added to Spirit Talker.