This app is a data capture tool for researchers and engineers working with 3D Gaussian Splatting (3DGS), NeRF, SLAM, and point cloud reconstruction
This app is a data capture tool for researchers and engineers working with 3D Gaussian Splatting (3DGS), NeRF, SLAM, and point cloud reconstruction. DESIGNED FOR DEVICES WITH A LiDAR SCANNER This app is designed for iPhone and iPad models equipped with a LiDAR scanner. Compatible devices include iPhone 12 Pro and later Pro models, and iPad Pro models with LiDAR. Some features are limited on devices without a LiDAR sensor. DATA CAPTURE ONLY — NOT A 3D RECONSTRUCTION APP This app does not perform 3D Gaussian Splatting generation, NeRF training, detailed point cloud creation, or SLAM processing. Transfer captured data to a computer and process it with external software such as NerfStudio, COLMAP, or other tools. Captures precisely synchronized sensor data from the LiDAR scanner, camera, and IMU on a per-frame basis, and exports it in formats ready for downstream 3D reconstruction pipelines. [Synchronized Per-Frame Data] RGB color images (selectable from 640x480 to 1920x1440) LiDAR depth maps (256x192, 16-bit, 0.1 mm resolution) Depth confidence maps (256x192, low:0, medium:1, high:2) Camera intrinsics (scaled to selected resolution) 6DoF camera pose (4x4 transform matrix, quaternion, Euler angles) IMU data (accelerometer & gyroscope) Magnetic heading & calibration status Exposure metadata ARKit tracking quality [Export Formats] Per-frame JSON (full sensor metadata) Session JSON NerfStudio-compatible transforms.json (OPENCV camera model) — ready for ns-train Binary PLY mesh (ARKit Scene Reconstruction) Split ZIP export (large sessions are automatically split into 1,000-frame chunks) [Exporting & Transferring to a Computer] Export captured sessions as ZIP files and move them to your computer through the native iOS Files app: - Save to the Files app (iCloud Drive or On My iPhone/iPad) - Write directly to a USB drive or external SSD, then connect it to your PC - Share via AirDrop or cloud storage Once unzipped on your computer, just point NerfStudio to the transforms.json and frames/ folder to start training (e.g., ns-train nerfacto). [Real-Time Capture Assistance] During scanning, an AR mesh overlay shows the areas already covered, giving visual feedback of scanning completeness. In addition, the app detects quality-degrading conditions — moving too fast, low light, or getting too close — and shows on-screen warnings, helping you collect uniform, high-quality data. [Key Features] Selectable capture rate: 5 / 10 / 15 / 20 / 25 / 30 fps Real-time AR mesh overlay during scanning (visual coverage feedback) Real-time capture guidance (warnings for fast motion, low light, too close) Exposure & white balance lock (keeps color and brightness consistent across frames) Selectable depth source (Raw / Smoothed) Post-capture quality summary (frame count, tracking quality, warning counts, etc.) Session naming and memo notes Settings screen (FPS, resolution, and options in one place) Automatic safe stop when free storage falls below 1 GB Orientation lock during capture Session management with thumbnails
| 序号 | 名称 | 价格 | 货币 | 地区 |
|---|---|---|---|---|
| 1 | 项目 | ¥ 68 | CNY | 中国 |
