Understanding the LCC Studio Pipeline
How LCC Studio converts XGRIDS scan data into 3D Gaussian Splatting models through SLAM optimization, Gaussian reconstruction, and camera coloring synchronization.
What LCC Studio Does
LCC Studio (Lixel CyberColor Studio) is the 3D Gaussian Splatting processing software for the XGRIDS ecosystem. It accepts raw scan data from the L2 Pro, K2, K1, and PortalCam, and produces photorealistic 3D models in proprietary and open formats optimized for visual fidelity, compact file size, and web sharing.
The output is not a point cloud. It is a Gaussian Splatting model, a scene represented as millions of oriented, colored ellipsoids that collectively produce a photorealistic appearance under lighting. Gaussian models excel at visual quality and efficient rendering for visualization, virtual tours, and web sharing. They are not the correct format for dimensional measurement, BIM modeling, or GIS workflows. Those use cases require LixelStudio and point cloud output.
LCC Studio provides four reconstruction modes: Single Scene for individual scan sessions, Map Fusion for combining multiple ground scan segments, Aerial-Ground Map Fusion for merging drone aerial imagery with ground scans, and Aerial Reconstruction for processing drone imagery without any ground scan data. The first three modes require data from a Lixel device. Aerial Reconstruction processes 100 to 10,000 drone images directly with no Lixel device required.
Device support in LCC Studio: LCC Studio supports L2 Pro, K2, K1, and PortalCam. LixelStudio supports L2 Pro, K2, and K1 only. If you are working with PortalCam data, LCC Studio is the only processing path available. K2 in LCC Studio is restricted to Single Model reconstruction (with Spatial Recognition supported as an Advanced Feature) and is excluded from Map Fusion, Aerial-Ground Fusion, and HD Enhancement.
Pipeline Stages
Stage 1: SLAM Optimization
Like LixelStudio, LCC Studio begins by reconstructing the device trajectory from raw LiDAR and IMU data. The SLAM result determines the geometric accuracy of the model. LCC Studio provides four Special SLAM Mode options to match scan conditions.
Auto (recommended): Intelligently selects the best reconstruction strategy. Prioritizes high-accuracy mode first; if vibration or interference causes failure, automatically switches to Robust mode and retries. Suitable for most scenarios.
None: Prioritizes high modeling accuracy. Suitable for stable devices and clear environments. Capture with vibration or interference may cause reconstruction failure.
Robust (default): Balances accuracy and stability with moderate interference tolerance during capture. Suitable for most scenarios.
Narrow Scene: Optimized for tunnels, mine shafts, long corridors, and other narrow environments. Using this in normal scenarios may cause failure.
Stage 2: Gaussian Splatting Reconstruction
LCC Studio converts the SLAM-derived geometry and camera image data into a 3DGS model, a scene represented as millions of oriented ellipsoids, each with position, color, opacity, and covariance parameters. This stage is what makes LCC Studio output fundamentally different from point cloud output.
The visual quality of the final model is determined by the reconstruction quality setting (Fast, Standard, Slow), the Max Gaussian Splats limit (typically ≤ 25M in Single Model mode), and the quality of the camera data collected during the scan.
Stage 3: Camera Coloring Synchronization
Camera images recorded during the scan are synchronized with the SLAM trajectory and used to color the Gaussian splats. This is equivalent to the coloring stage in LixelStudio but operates on Gaussian geometry rather than point geometry.
Each XGRIDS device provides its own camera data without external accessories. The K2 produces real-time true-color point cloud rendering during scanning with imagery fused as you scan. The L2 Pro uses dual 48 MP panoramic cameras plus a visual positioning camera. The K1 uses dual 48 MP panoramic cameras. The PortalCam uses a four-camera array (two panoramic, one front, one macro) optimized for close-range fidelity.
The LCC Format
LCC is a proprietary compressed format for Gaussian Splatting models. It is 10 to 30% of the size of an equivalent PLY file, making it the preferred format for delivery and archiving within the XGRIDS ecosystem. Models in LCC format can be viewed in LCC Studio, shared via LCC's one-click web publish, and opened in the LCC for BIM plugin in Revit.
PLY is the standard open format for 3DGS models and offers maximum compatibility with third-party tools. PLY files are 5 to 10 times larger than equivalent LCC files for the same model. Use PLY when the recipient uses software outside the XGRIDS ecosystem.
LCC2 (next-generation, introduced). LCC2 offers smaller file sizes, smoother rendering, and faster model loading compared to LCC. Editing features (Selector, Crop, Color Grading) are not yet available for LCC2. LCC and LCC2 cannot be mixed within the same project. Use LCC2 when storage and rendering efficiency outweigh the need for in-Studio editing operations.
Hardware Requirements
LCC Studio is Windows only and requires an NVIDIA GPU with CUDA support. AMD and Intel discrete GPUs are completely unsupported with no workaround. The hardware threshold for usable processing is significantly higher than for LixelStudio.
Single Scene and General Use
Memory vs. data processing capabilities. 64 GB stable processing supports up to 30 minutes of capture data; processing more than 45 minutes may cause reconstruction failure. 128 GB stable processing supports up to 60 minutes of capture data; processing over 90 minutes increases reconstruction failure risk.
Map Fusion and Aerial-Ground Fusion
Map Fusion and Aerial-Ground Fusion place substantially higher demands on the processing computer than Single Model reconstruction. The hardware listed below is required for these workflows. Running fusion on hardware below these thresholds results in out-of-memory failures and corrupted or incomplete fusion outputs.
Large model recommendation. For models with total length 150 minutes or more requiring high-quality reconstruction, 96 to 128 GB memory is recommended. If memory is insufficient, select Standard quality to ensure smooth processing.
©2026 Alpine Reality Capture LLC • XGRIDS Pro Guide™ • Site Disclaimer