.png)
.png)
.png)
Without an available network, login may not complete, some advanced functions such as 3DGS may not work normally, and logs cannot be uploaded for technical support.
Size the Workstation for Data Volume and Processing Tasks
Workstation requirements primarily depend on data volume, processing stage, and the number of concurrent tasks—not just the device model. Evaluate the central processing unit (CPU), system memory, the graphics processing unit (GPU), video memory (VRAM), storage, drivers, and whether the intended workflow includes point-cloud mapping, colorization, or 3D Gaussian splatting (3DGS). Resource consumption can differ between a small-room project and a multi-floor project captured with the same model.
SHARE PointClouds Studio V2.6 recommends 64-bit Windows 11, an Intel Core i7-10700H at 2.90 GHz or an AMD equivalent, at least 64 GB of memory, at least 100 GB of available storage, and an NVIDIA RTX 3060 or higher-specification graphics card. Point-cloud mapping requires at least 8 GB of dedicated VRAM, while 3DGS requires at least 12 GB of dedicated VRAM. When NVIDIA parallel computing platform (CUDA) acceleration is used for point-cloud colorization, the NVIDIA driver should be version 572.83 or later. Some Windows 10 environments may encounter issues; test with a non-critical sample before upgrading. These requirements apply when planning processing environments for data from the SHARE SLAM S20, SHARE SLAM S20 SE, SHARE SLAM S100, and the SHARE3DCAM C10 Series. For actual capture capabilities and data volumes, refer to the model documentation and real project conditions.
Figure 1: Different computing resources affect different processing stages. If performance is sluggish, first reduce the task scope, then check the environment and hardware in sequence.
How CPU, GPU, Memory, and Storage Affect Point-Cloud Processing
| Resource | Primary impact | Common symptoms when insufficient |
|---|---|---|
| CPU | Point-cloud and 3DGS computation, registration, and coordinate calculations | Longer computation or processing times |
| GPU/VRAM | Point-cloud rendering, CUDA colorization, and 3DGS | Sluggish viewing, slower tasks, or task failure |
| Storage | Reading and writing raw data, projects, photos, and deliverables | Insufficient space, slower read/write operations, or task failure |
| Network | Sign-in, updates, and log uploads | Network-dependent steps cannot be completed; local processing does not mean cloud processing |
Optimize the Workflow Before Deciding Whether to Upgrade Hardware
- Record the project file size, approximate point count, photo volume, processing stage, and number of concurrent tasks.
- Preserve the complete raw data, and use 2D or 3D cropping to exclude irrelevant areas from the current task.
- Close unused point clouds and applications, reduce concurrent tasks, and check data size, disk space, and memory use before processing.
- Check the full software version, graphics driver, operating-system version, data size, memory use, and remaining disk space before deciding whether to expand hardware capacity. Keep the complete raw data and verified deliverables backed up separately.
Resource and Quality Boundaries Across Processing Stages
Manual corresponding-point registration: Select fixed corresponding features between the reference point cloud and the point cloud to be registered.
The result is saved as original_name_registered.las. Higher-specification hardware can improve interaction and processing performance, but it cannot compensate for incorrect or poorly distributed corresponding points.
- Coordinate transformation: Calculate parameters after providing at least three valid control points, or enter verified XYZ offsets, rotations, and scale. The result is saved as
original_name_converted.las. Better performance does not mean better coordinate quality. - External post-processed kinematic (PPK) trajectory input for global navigation satellite system (GNSS)-aided processing: Use an external PPK tool to generate an RTKLIB POS trajectory file, inspect its trajectory, and select the file in SHARE PointClouds Studio only as the
Trajectory SourceinGNSS Fusion. Computer specifications cannot substitute for base-station data, time synchronization, or trajectory quality. - Linked trajectory and photo viewing: Photo comparison requires undistorted photos. Trajectories, photo locations, photos, and a floor plan or 3D viewport can be viewed together. Photo completeness and disk speed both affect the experience.
- 3D/elevation accuracy report: The two checks should be performed independently and recorded separately. Accuracy cannot be judged from rendering smoothness.
- Computer-aided design (CAD)-assisted drafting: CAD drawing export simultaneously generates drawing exchange format (DXF) and DWG files and includes them in the exported ZIP package. Section or plan-view deliverables must be manually corrected and reviewed; RVT, IFC, and complete BIM models remain downstream professional workflows.
- 3DGS: Intended for photorealistic viewing, not measurement acceptance. Increasing graphics-card specifications or iteration counts does not turn a visual deliverable into a measurement deliverable.
Troubleshooting Order for Sluggish Performance or Failures
First verify the data volume and number of concurrent tasks, then check the software version, driver, operating system, memory, VRAM, and disk. Failed or canceled reconstruction tasks can be retried in Task Management. Do not repeatedly submit the same large task before identifying the cause.
If the issue persists, retain the project, task status, and logs. You must sign in before uploading logs. Support materials should include the software version, operating system, CPU, memory, graphics card, data volume, reproduction steps, error screenshots, and a sanitized sample.
FAQ
Q: Is a GPU mandatory? Not all viewing and management operations depend equally on the GPU, but point-cloud rendering, CUDA colorization, and 3DGS make significant use of it. SHARE PointClouds Studio V2.6 recommends an RTX 3060 or higher, with at least 8 GB of dedicated VRAM for point-cloud mapping and at least 12 GB for 3DGS.
Q: Is 64 GB of memory guaranteed to be enough for every large S100 project? No. Although 64 GB is the current recommendation, actual requirements also depend on point count, photos, partitioning, and concurrent tasks. Large projects should first be tested with representative data.
Q: Will upgrading the graphics card improve registration or coordinate accuracy? Not directly. Registration depends on the selection of corresponding points, while coordinate transformation depends on control points or verified parameters. A 3D/elevation accuracy report is still required.
Q: Why is 3DGS more sensitive to VRAM, photo quality, and scene scale? 3DGS processes imagery and additional graphics data, so its memory, image-quality, and scene-size conditions differ from point-cloud viewing. Validate representative data and divide large scenes where appropriate. 3DGS is intended for viewing, not measurement.
Q: What should I submit after a task fails? First preserve the task status and logs, then check data volume, software version, and workstation resources. Retry only after confirming that the inputs are complete and the documented resource conditions are met. If the issue remains, submit the software version, system and hardware information, data volume, reproduction steps, screenshots, logs, and a sanitized sample.
