Section 1: Understanding the XGRIDS Ecosystem
Read this section first, regardless of which device you own. Every decision you will make about hardware, workflow, and software connects to the foundational concepts covered here.
Quick Reference
XGRIDS Product Line at a Glance
The L2 Pro family shares the same physical platform — 1.7 kg, 1TB SSD, 90-minute battery, IP54, ≤3.0 cm absolute accuracy — and differs only in LiDAR configuration. The table below shows where the variants diverge.
L2 Pro — Model Comparison
| Spec | 16-120 | 32-120 | 32-300 |
|---|---|---|---|
| LiDAR Channels | 16 | 32 | 32 |
| Working Range | 0.5m to 120m | 0.5m to 120m | 0.5m to 300m |
| Scan Speed | 320,000 pts/s | 640,000 pts/s | 640,000 pts/s |
| Best For | Frugal AEC projects | Professional AEC work | Tunnels, large outdoor sites, tall structures |
Device Family Comparison
| Spec | L2 Pro (any model) | K1 | PortalCam |
|---|---|---|---|
| Working Range | 0.5m to 300m (model dependent) | 0.1m to 40m | 0.1m to 60m |
| Scan Speed | 320K–640K pts/s | 200,000 pts/s | 850,000 pts/s |
| Weight | 1.7 kg | 1.0 kg | 870 g |
| Battery Life | 90 min | 90 min | 60 min |
| Storage | 1TB SSD | 256GB TF (swappable) | 512GB EMMC |
| Relative Accuracy | ≤1.2 cm | ≤1.2 cm | 2 cm RMSE |
| Absolute Accuracy | ≤3.0 cm | ≤3.0 cm | — |
| Primary Output | Point cloud | Point cloud | 3DGS |
| Processing Software | LixelStudio | LixelStudio/LCC Studio | LCC Studio |
Software Ecosystem Map
| Software | Purpose | Works With | Cost |
|---|---|---|---|
| LixelGO | Mobile scan controller | L2 Pro, K1 | Free |
| LCC Scan | Mobile scan controller | PortalCam | Free |
| LixelStudio | Point cloud processing | L2 Pro, K1 | Free (included) |
| LCC Studio | 3DGS reconstruction | L2 Pro, K1, PortalCam | $1,500/year |
| LixelWeb | Browser sharing | All devices | Included with LCC Studio |
Processing Pipeline Decision
| If you need... | Use this pipeline |
|---|---|
| Survey-grade measurements | LixelStudio (point cloud) |
| CAD / BIM deliverables | LixelStudio (point cloud) |
| Client-facing visualization | LCC Studio (3DGS) |
| Web-shareable 3D model | LCC Studio (3DGS) |
| Virtual tours or walkthroughs | LCC Studio (3DGS) |
| Measurements & visualization | Run both pipelines from the same raw scan |
Computer Requirements Summary
LixelStudio
| Spec | Minimum | Recommended |
|---|---|---|
| CPU | Intel i7 11th Gen | Intel i9 12th Gen |
| GPU | NVIDIA RTX 3060 | NVIDIA RTX 3070 or better |
| RAM | 32 GB | 64 GB |
| Storage | 500 GB | 2 TB SSD |
LCC Studio (NVIDIA GPU required — AMD not supported)
| Spec | Recommended | Optimal |
|---|---|---|
| CPU | 16-core (e.g. AMD Ryzen 9 9950X) | Same |
| GPU | NVIDIA RTX 3090 (24 GB VRAM) | NVIDIA RTX 4090 (24 GB VRAM) |
| RAM | 64 GB DDR5 | 96–128 GB (scans over 150 min) |
| Storage | 2 TB+ SSD | Same |
1.1 — What Is XGRIDS Reality Capture?
XGRIDS makes mobile scanning systems that produce precise 3D representations of physical environments. Unlike a camera that records light, or a tape measure that records a single distance, a mobile scanner captures millions of precise spatial measurements per second while you walk through a space. The result is a dense, three-dimensional record of everything the scanner could see — walls, floors, ceilings, structural elements, equipment, and outdoor terrain — all in their correct spatial relationships to each other.
The word "mobile" is what distinguishes this category from the tripod-based terrestrial laser scanners that have traditionally dominated professional 3D capture work. A tripod scanner is placed at a fixed position, captures a single station of data, and then must be repositioned to the next station. Mobile scanning removes that constraint entirely. You carry the scanner through a space continuously, and it builds the 3D model as you move.
SLAM: The Core Technology
What makes continuous mobile scanning possible is a technology called SLAM, which stands for Simultaneous Localization and Mapping. This is worth understanding because it affects everything about how you should operate an XGRIDS device — why you move the way you move, why loop closures matter, and why certain environments are more challenging than others.
A SLAM system solves a problem that sounds simple but is computationally intensive: to build a map, you need to know where you are, but to know where you are, you need a map. XGRIDS devices solve this with a multi-sensor approach that combines three data streams simultaneously. The LiDAR (Light Detection and Ranging) sensor fires laser pulses in all directions and measures the time it takes for each pulse to return, which gives precise distance measurements to every surface. The IMU (Inertial Measurement Unit) tracks acceleration and rotation, so the system always knows how the device is moving between LiDAR pulses. The visual cameras track recognizable features in the environment and use them as spatial anchors. The SLAM software fuses all three inputs continuously to estimate the scanner's position at every moment and build the point cloud relative to that position.
The practical implication of this architecture is that SLAM performance depends on the environment providing trackable features. Open spaces with few distinctive surfaces — long blank corridors, large empty rooms, open outdoor fields — give the SLAM system fewer anchors to work with, which can lead to drift. Understanding this is fundamental to scanning technique, which is covered in Section 4.
LiDAR + Camera Integration
Every XGRIDS device captures two things simultaneously: precise geometry from LiDAR, and color information from cameras. These are registered together so that each point in the final output has both a precise 3D position and accurate color information. This integration is what allows XGRIDS outputs to look like photorealistic scenes rather than the abstract blue-and-green point clouds associated with older scanning technology.
The two software pipelines handle this integration differently. LixelStudio produces a colored point cloud — a set of individual measured points, each with an XYZ position and an RGB color value. LCC Studio produces a 3D Gaussian Splatting model — a representation built from millions of oriented, colored ellipsoids that together create a photorealistic visual output. Both start from the same sensor data, but the outputs serve different professional purposes, which Section 1.4 covers in detail.
Troubleshooting 1.1
Problem: The scanner loses track or produces obviously distorted geometry in certain areas. This is a SLAM tracking failure, and its most common cause is an environment with insufficient visual and geometric features. Blank white walls, uniform carpet, and large open outdoor spaces all present this challenge. The solution is scanning technique rather than a software setting: move more slowly, vary your path to see surfaces from multiple angles, create loop closures by returning to previously scanned areas, and add temporary physical markers or objects to featureless surfaces if the project allows. Section 4 covers this in depth.
1.2 — The Hardware Family
XGRIDS produces three scanning devices, each representing a distinct design philosophy and serving different professional use cases. Understanding where each device excels — and where it falls short — is the foundation for making the right equipment decision for any given project.
Lixel L2 Pro
The L2 Pro is XGRIDS's industrial-grade scanner, built for professionals who need high accuracy across large and complex sites. It is heavier and more expensive than the K1, but it offers significantly greater range, faster scan speeds, and better performance in challenging outdoor and large-scale environments. The L2 Pro is available in three configurations based on LiDAR channel count and maximum working range.
The L2 Pro 16-120 has 16 LiDAR channels, a 120m working range, and a scan speed of 320,000 points per second. It is the entry point to the L2 Pro family and covers the majority of AEC use cases. The L2 Pro 32-120 doubles the LiDAR channels to 32 and the scan speed to 640,000 points per second while maintaining the same 120m range — this is the most commonly specified configuration for professional AEC work because it offers the best combination of point density and accuracy for building-scale projects. The L2 Pro 32-300 extends the working range to 300m, which makes it capable of capturing degradation scenarios like tunnels without drift and suitable for very large outdoor sites, infrastructure, and terrain; however, at distances beyond 120m, absolute accuracy is reduced compared to the 32-120 at closer range.
All three L2 Pro models share the same physical specifications: 1.7 kg without battery, 1TB onboard SSD, 90-minute battery life, and IP54 environmental protection. The panoramic camera system uses two 48MP sensors for a maximum panoramic image resolution of 56MP, and a separate 1MP camera handles visual positioning. The LiDAR field of view is 360° × 270°, giving thorough spatial coverage in all directions. Absolute accuracy is ≤3.0 cm and relative accuracy is ≤1.2 cm under the 3σ rule when operating with RTK. SLAM drift is specified at RMSE 3cm over 100m of travel.
The L2 Pro is the right device for campus-scale environments, large industrial facilities, multi-building sites, outdoor terrain, infrastructure documentation, and any project where the scan area spans tens of thousands of square meters on a single battery charge. It also supports drone-mounted operation on the DJI Matrice 300 RTK and Matrice 350 RTK, which is covered in Section 12.
Lixel K1
The K1 is designed for indoor environments and single building projects. At 1.0 kg and with a swappable 256GB TF card instead of a fixed SSD, it is lighter and more portable than the L2 Pro. The tradeoff is a shorter working range of 0.1m to 40m and a lower scan speed of 200,000 points per second — which is appropriate for the closer quarters and smaller volumes of indoor single-building work.
The K1 uses four cameras: two 12MP panoramic sensors, one 12MP front camera, and one 12MP macro camera. This four-camera system gives it strong visual positioning performance in the interior environments where it is designed to operate. The LiDAR field of view is 360° × -7~+52°. Absolute accuracy is ≤3.0 cm and relative accuracy is ≤1.2 cm, consistent with the L2 Pro. SLAM drift is specified at RMSE 3cm over 50m — half the distance of the L2 Pro, reflecting the K1's shorter-range design envelope.
Battery life is 90 minutes, matching the L2 Pro, but the swappable TF card means storage can be extended in the field without returning to a workstation, which is a practical advantage on multi-room projects where session management requires flexibility. The K1 is ideal for scanning single low-rise buildings, both indoor and outdoor coverage of building-scale projects, and areas covering several thousand square meters on a single battery.
The K1 can process data through both LixelStudio (point cloud) and LCC Studio (3DGS), making it the most flexible device in the lineup for teams that need to serve both survey and visualization deliverables from the same hardware investment.
PortalCam
The PortalCam is a fundamentally different device from the L2 Pro and K1. Where those two scanners are primarily designed for accuracy and range, the PortalCam is optimized for the highest-quality 3D Gaussian Splatting output — it is the device to choose when photorealistic visual quality is the primary requirement.
It is the smallest and lightest of the three, at 870g without the tripod and 130 × 90 × 77mm in body dimensions. Despite its size, it houses a 96-channel LiDAR sensor — the highest channel count in the lineup — at 850,000 points per second scan speed. The LiDAR field of view is a full 360° × -90~+90°, capturing a complete hemisphere in all directions. Working range is 0.1m to 30m at 10% reflectivity and 0.1m to 60m at 90% reflectivity. Relative accuracy is 2cm RMSE and repeat accuracy is 5cm RMSE.
The PortalCam uses a tripod-mounted stationary scanning method rather than handheld walking, which produces a different type of motion profile for the SLAM algorithm. It is paired exclusively with LCC Studio and the LCC Scan mobile app. It does not produce point cloud output compatible with LixelStudio. Battery life is 60 minutes — shorter than the other two devices — making battery management a more active consideration on longer projects.
The PortalCam also supports the full range of LCC Studio advanced features including Map Fusion, Aerial-Ground Map Fusion, HD Enhancement, and Spatial Recognition. It was designed around the LCC Studio workflow, and its capabilities are most fully realized in that context.
Choosing the Right Device
The decision between devices comes down to three questions: what type of output does the deliverable require, what is the scale of the site, and what is the primary environment?
If the deliverable is a point cloud for survey, CAD, or BIM work, the choice is between the L2 Pro and K1. For building-scale interior work, the K1 is sufficient and more portable. For multi-building sites, large facilities, outdoor terrain, or any project where scan area exceeds what the K1 can cover in a reasonable session count, the L2 Pro is the right choice. The L2 Pro 32-120 covers the majority of professional AEC requirements.
If the deliverable is a 3DGS model for visualization or web delivery, any of the three devices can be used with LCC Studio. The PortalCam will produce the highest visual quality output because it was designed for that purpose. The L2 Pro and K1 produce strong results as well, particularly in larger spaces where the PortalCam's shorter range would require more scan positions to achieve full coverage.
If both deliverable types are required on the same project, the L2 Pro or K1 is the practical choice because it can feed both pipelines from a single scan session. The PortalCam cannot produce LixelStudio point cloud output.
Troubleshooting 1.2
Problem: Unsure whether the L2 Pro 32-120 or 32-300 is the right choice for a specific project. The 32-300's extended range primarily benefits two scenarios: outdoor sites where the scanner needs to capture structures or terrain beyond 120m, and environments with degraded SLAM conditions (such as long tunnels or open industrial spaces) where the longer range helps maintain tracking stability. For standard AEC building work, the 32-120 is more accurate at the distances where most useful geometry is captured. If the tallest structure on site is under 120m and the project is primarily building-focused, the 32-120 is the better specification.
Problem: The K1 is running out of storage before the scan session is complete. The K1's 256GB TF card is swappable in the field. Carry at least one spare formatted card for longer projects. Before each project, calculate the expected data volume based on the scan area and duration, and plan the number of scan sessions and card swaps accordingly. Typical K1 data rates produce approximately 1.5 to 3 GB per minute of scanning depending on environment complexity.
1.3 — The Software Ecosystem
XGRIDS hardware pairs with a set of software tools that cover the full workflow from scanning to delivery. Each piece of software has a specific role, and understanding which tool does what prevents confusion about where to go at each stage of a project.
LixelGO (Mobile App — L2 Pro and K1)
LixelGO is the mobile app that controls the L2 Pro and K1 during scanning. It connects to the scanner over WiFi and serves as the operational interface for starting and stopping scans, monitoring the real-time point cloud preview, configuring RTK settings, setting mount type for drone mode, managing scan sessions, and reviewing data before transfer. LixelGO is available for both iOS and Android.
Critically, LixelGO is not just a convenience — for certain configurations like drone mode, it is the only correct way to start a scan, because it writes mount type and RTK configuration into the project file that LixelStudio reads during processing. If LixelGO is bypassed, that metadata is not recorded and the project cannot be processed correctly.
LCC Scan (Mobile App — PortalCam)
LCC Scan is the PortalCam's dedicated control app, analogous to LixelGO but built specifically for the PortalCam workflow. It handles device connection, scanning sessions, control point marking for Map Fusion and Aerial-Ground Map Fusion, and real-time preview. It connects via Bluetooth for initial pairing and WiFi for ongoing operation. LCC Scan is available for iOS and Android.
LixelStudio (Desktop — Point Cloud Processing)
LixelStudio is the desktop software that processes raw L2 Pro and K1 scan data into finished point clouds. It is free and included with the scanner. LixelStudio handles SLAM trajectory optimization, point cloud generation, coloring from camera data, Map Fusion for stitching multiple sessions, RTK and ground control point integration for absolute accuracy, and export to industry-standard formats.
LixelStudio exports point clouds in the following formats: LAS and LAZ (for GIS tools and most point cloud workflows), E57 structured (compatible with Revit, AutoCAD, and Navisworks), OBJ with texture, and RCP/RCS for the Autodesk ecosystem. It runs on Windows and requires an NVIDIA GPU. The minimum configuration is an Intel i7 11th generation processor, an RTX 3060, and 32 GB of RAM. The recommended configuration is an i9 12th generation, an RTX 3070 or better, and 64 GB of RAM with 2 TB of storage.
LCC Studio (Desktop — 3D Gaussian Splatting)
LCC Studio is the desktop software that produces 3D Gaussian Splatting models from Lixel scanner data. It is a subscription product at $1,500 per year. LCC Studio accepts raw data from the L2 Pro, K1, and PortalCam and produces photorealistic 3DGS outputs in .lcc (proprietary compressed format, 70-90% smaller than PLY), .ply (standard 3DGS format, compatible with UE/Unity), .usdz, and 3D Tiles. It also handles Map Fusion, Aerial-Ground Map Fusion, Aerial Reconstruction, HD Enhancement, and Spatial Recognition — all of which are advanced features covered in later sections of this guide.
LCC Studio requires a CUDA-capable NVIDIA GPU. AMD graphics cards are not supported. The recommended processing configuration is a 16-core AMD Ryzen 9 9950X or equivalent processor, 64 GB DDR5 RAM (96-128 GB recommended for scan sessions over 150 minutes in length), and an RTX 3090 or RTX 4090. Under recommended settings at Medium quality, LCC Studio processes approximately 1 minute of scan data every 20 minutes, producing an approximate 1:20 ratio of scan time to processing time.
Basic vs. Premium capabilities: The standard LCC Studio subscription includes single-scene reconstruction, model viewing, editing, and web publishing. Premium features — Map Fusion, Aerial-Ground Map Fusion, Aerial Reconstruction, HD Enhancement, and Spatial Recognition — require a higher-tier subscription. Verify your subscription level before planning a project that depends on these features.
LixelWeb
LixelWeb is the browser-based platform for sharing and viewing LCC models. When you publish a model from LCC Studio, it becomes accessible via a web link that clients and collaborators can open in any modern browser on desktop, tablet, or mobile without installing any software. LixelWeb supports password-protected sharing, public links, and embedded viewers. It is included with the LCC Studio subscription.
The Two Parallel Pipelines
One of the most important concepts in the XGRIDS ecosystem is that LixelStudio and LCC Studio represent two completely separate processing pipelines. They take the same raw scan data as input, but they produce fundamentally different outputs using different algorithms, and there is no conversion path between them.
The raw data from a single scan session can be processed through LixelStudio to produce a point cloud, and that same raw data can also be processed through LCC Studio to produce a 3DGS model. These two operations are independent — running one does not affect the other, and the outputs cannot be converted between formats. A LixelStudio point cloud cannot be converted into a 3DGS model. A 3DGS model from LCC Studio cannot be converted into a survey-grade point cloud.
This matters practically because it determines what you need to collect in the field. If a project will require both a point cloud and a 3DGS model as deliverables, the raw scan data needs to be retained and transferred for both processing runs. The field collection is the same — the data just gets processed twice, once in each pipeline.
Troubleshooting 1.3
Problem: LixelStudio will not open or crashes immediately on launch. LixelStudio requires an NVIDIA GPU. If the workstation uses an AMD or Intel integrated GPU, LixelStudio will not run. Verify the GPU in Device Manager before installation. If an NVIDIA GPU is present but the software still fails to launch, update the NVIDIA driver to the latest stable release and then reinstall LixelStudio.
Problem: LCC Studio processing fails with a CUDA error. LCC Studio requires CUDA, which is exclusive to NVIDIA hardware. An AMD GPU of any specification will not work with LCC Studio. If the machine has an NVIDIA GPU and CUDA errors are still occurring, confirm that the NVIDIA driver is current and that the VRAM meets the minimum requirement for the project size. Processing very large scenes on a GPU with limited VRAM (under 12 GB) is the most common source of CUDA memory errors.
Problem: A client is asking for a point cloud but the project was only processed through LCC Studio. LCC Studio produces 3DGS output in .lcc and .ply formats. These are not survey-grade point clouds. The .ply output can be opened in some point cloud tools, but it does not contain the accuracy characteristics of a LixelStudio-processed point cloud. If a survey-grade point cloud is required and the raw scan data still exists, process it through LixelStudio. If the raw data was not retained, the project must be rescanned. This situation reinforces why it is important to understand the deliverable requirements before processing, and to retain raw scan data until all deliverables have been confirmed as accepted.
1.4 — Output Types Explained
The two output types the XGRIDS ecosystem produces — point clouds and 3D Gaussian Splatting models — serve different professional purposes and reach different audiences. Choosing the wrong output type for a given deliverable is one of the most common sources of client confusion and project rework.
Point Clouds
A point cloud is a collection of individual measured points in three-dimensional space. Each point has a precise XYZ coordinate, and in a colored point cloud, each point also has RGB color values derived from the scanner's cameras. The defining characteristic of a point cloud is that it consists of actual measurements — each point represents a specific location in the real world that the LiDAR sensor detected, with a spatial accuracy governed by the scanner's specifications.
Point clouds are the standard deliverable for workflows that require measurements: cross-sections, floor-to-ceiling heights, structural dimensions, deviation analysis, and as-built documentation for CAD or BIM. Revit, AutoCAD, Navisworks, and most other professional AEC tools import point clouds directly through standardized formats like E57, LAS, and RCP. A surveyor, a structural engineer, or a MEP designer working in Revit is expecting a point cloud.
The limitation of point clouds is visual density. Even a high-density point cloud looks like a field of individual dots at close range. It communicates precise spatial information effectively to professionals trained to read it, but it is not inherently photorealistic, and it requires specialized software to view properly. It is not the right format for client presentations, marketing, or non-technical stakeholders.
3D Gaussian Splatting
3D Gaussian Splatting (3DGS) is a rendering method that represents a scene using millions of small, oriented, semi-transparent ellipsoids — called Gaussians — each with a position, shape, opacity, and color. When rendered together, these Gaussians produce a photorealistic image that looks like a photograph of the scanned space from any viewpoint. The visual result is much closer to a photograph or video walkthrough than a traditional point cloud.
The key advantage of 3DGS for client-facing work is accessibility. An LCC model published to LixelWeb can be explored by anyone with a web browser and a link — no software installation, no specialized training, no understanding of point cloud visualization. A client can walk through their facility virtually on a tablet the day after scanning, which is a substantially different capability than delivering an E57 file that requires Revit or Navisworks to open.
The tradeoff is that 3DGS is a rendering format, not a measurement format. While LCC Studio includes measurement tools, the measurements derive from the underlying LiDAR data, and the visual representation is not the same as a survey-grade point cloud. For deliverables where measurements will be taken and used for design, construction, or legal purposes, a point cloud from LixelStudio is the authoritative output.
When to Use Each
The question of which output to use is ultimately determined by who will receive the deliverable and what they will do with it. For a structural engineer creating a model in Revit, the answer is a point cloud. For a facilities manager reviewing as-built conditions in a browser, the answer is 3DGS. For a construction client reviewing progress against design, 3DGS is more useful. For a surveyor performing deviation analysis, the point cloud is essential.
Many projects benefit from both. The raw scan data can be processed through both pipelines, producing a point cloud for technical documentation and a 3DGS model for stakeholder communication — two different deliverables from a single field collection session. Section 1.3 explains why this requires retaining the raw data for both processing runs.
Troubleshooting 1.4
Problem: A client received a .ply file and cannot open it. The .ply format requires a viewer that supports 3DGS. Standard 3D software tools like Blender or MeshLab can open PLY files but will display them as point geometry rather than as rendered Gaussians. For clients without technical software, the correct delivery method is a web link via LixelWeb, which requires no software installation on the client's side. Publish the model from LCC Studio and share the link directly.
Problem: A client is asking for measurements from a 3DGS model that has already been delivered. Measurements can be taken within LCC Studio's viewer tools. Distance and area measurements are available and derive from the underlying LiDAR data. Share the measurement workflow with the client if they have LCC Studio access, or take the requested measurements yourself and deliver them as annotated screenshots or a PDF report. If the project requires extensive measurement extraction, that work should be planned as part of the deliverable scope rather than added after delivery.
1.5 — Computer Hardware Requirements
Processing 3D scan data is among the most computationally intensive tasks a workstation will perform. The hardware requirements for XGRIDS processing are not suggestions — running LixelStudio or LCC Studio on undersized hardware produces slower processing, more failures, and in some cases results that are never completed at all. Getting the hardware right before a project begins is far less expensive than failed processing runs on deadline.
LixelStudio Requirements
LixelStudio processes point clouds on the CPU with GPU acceleration. It runs on Windows and requires an NVIDIA GPU, though the VRAM requirement is lower than LCC Studio because point cloud processing is less GPU-memory-intensive than 3DGS reconstruction.
The minimum configuration — an Intel i7 11th generation, RTX 3060, 32 GB RAM, and 500 GB storage — is sufficient for single-session projects of modest size. For multi-session projects with Map Fusion, longer scan durations, or high-density outdoor scans, the minimum configuration will produce very slow results and may run out of memory during processing. The recommended configuration of an i9 12th generation or better, an RTX 3070 or better, 64 GB RAM, and 2 TB of SSD storage handles professional-scale projects reliably.
Storage is a frequently underestimated requirement. Raw scan data, intermediate processing files, and final exports together can easily consume several hundred gigabytes per project. A 2TB SSD dedicated to project data (separate from the operating system drive) is a practical minimum for active professional use.
LCC Studio Requirements
LCC Studio processes 3DGS models using CUDA on the GPU. This means NVIDIA hardware is not just recommended — it is required. There is no AMD GPU support for LCC Studio processing. The GPU's VRAM is the most critical specification because the Gaussian Splatting reconstruction holds a significant portion of the model in VRAM during processing.
The recommended GPU is the RTX 3090 (24 GB VRAM) or RTX 4090 (24 GB VRAM). A consumer GPU with 12 GB VRAM such as the RTX 3060 can process shorter, simpler scenes but will fail on larger projects with the Insufficient Memory error. For projects using Map Fusion or Aerial-Ground Map Fusion — which combine multiple scan sessions into a single reconstruction — 24 GB VRAM is effectively a requirement rather than a recommendation.
For CPU and RAM, the recommended configuration is a 16-core processor (AMD Ryzen 9 9950X or equivalent) and 64 GB DDR5. For scan sessions over 150 minutes in total duration, 96 to 128 GB of RAM is recommended. Processing time scales approximately linearly with scan duration: at Medium quality on recommended hardware, expect roughly 20 minutes of processing time per 1 minute of collected scan data.
LCC Studio is distributed as an offline ISO installer as of version 1.9.1. Unlike the previous web-based installer, the ISO contains all dependencies and does not require an internet connection during installation.
Laptop vs. Desktop
A desktop workstation is strongly preferable for primary LCC Studio processing. The RTX 4090 is a desktop GPU — laptop variants of high-end NVIDIA cards have reduced VRAM and lower sustained performance under the continuous load of a multi-hour reconstruction. If field-deployable processing is necessary, a laptop with an RTX 4080Ti (16 GB VRAM) or RTX 4090 mobile will handle moderate projects, but its results will be slower and its maximum project size will be smaller than a comparable desktop.
LixelStudio is less demanding and runs acceptably on a capable laptop, making it more suitable for field use or on-site data review.
Storage Planning
The most common storage-related mistake is not accounting for the full project lifecycle: raw scan data from the device, working copies used during processing, temporary files generated during processing (which can reach 5 times the size of the input data), and final exports. For a project with 60 minutes of total scan time, plan for at least 200 to 400 GB of working space during LCC Studio processing, plus the final deliverable files.
Maintain raw scan data in a separate archive location until all deliverables have been confirmed and accepted. Processing failures requiring a rerun are common enough that deleting raw data before delivery confirmation is an unnecessary risk.
Troubleshooting 1.5
Problem: LCC Studio crashes during processing with an Insufficient Memory error. The most immediate intervention is to enable Low-Memory Mode in LCC Studio and reduce the Maximum Gaussian Splats slider. Closing all other applications before starting processing frees RAM and VRAM that other programs are holding. If the error persists even at reduced settings, the project may genuinely exceed what the hardware can handle. Splitting the project into smaller segments — processing them individually and then using Map Fusion — reduces the peak memory requirement at any single processing step.
Problem: Processing completes but takes far longer than expected. Verify that LCC Studio is using the NVIDIA GPU for processing and not falling back to CPU. Open the NVIDIA control panel and confirm that LCC Studio is assigned to the high-performance GPU rather than any integrated Intel or AMD graphics. On laptops, power settings can cause the GPU to throttle under sustained load — run LCC Studio on AC power with the Windows power plan set to High Performance.
Problem: The workstation has sufficient RAM and GPU but processing still runs slowly or fails. Check that the project path in LCC Studio is set to an SSD rather than a mechanical hard drive. LCC Studio writes large temporary files during processing, and a mechanical drive creates a severe bottleneck that can make processing times five to ten times longer than on SSD and can cause timeouts that result in processing failures. This setting is configured in LCC Studio's Settings menu before starting any project, and it is one of the first things to configure on a new installation.
