How to Clean Noisy Point Clouds in Metashape (Step-by-Step Guide)
News

How to Clean Noisy Point Clouds in Metashape (Step-by-Step Guide)

Posted on by admin

When working with Agisoft Metashape, one of the most common issues after generating a dense point cloud is the presence of noise — unwanted points that can degrade model accuracy and appearance. Whether you’re mapping urban environments or natural terrain, learning to clean noisy point clouds in Metashape is essential for creating high-quality outputs.

What Causes Noisy Point Clouds?

Noisy data in a dense cloud can result from various factors such as:

  • Poor image quality (motion blur, low contrast)
  • Shiny or reflective surfaces (glass, metal, water)
  • Low overlap or gaps in coverage
  • Vegetation movement (trees, grass)
  • Incorrect depth settings during processing

Why Clean Your Point Cloud?

A clean point cloud helps improve:

  • Accuracy of mesh and textured model
  • DSM/DTM calculations
  • Exported CAD and BIM workflows
  • Visualization clarity

Step-by-Step: Cleaning Point Clouds in Metashape

Here’s how to clean your dense cloud in Agisoft Metashape (v2.0+):

Step 1: Open the Dense Cloud

Once you’ve aligned the photos and generated the dense point cloud:

  • Switch to the “Model” view
  • Select the “Dense Cloud” layer in the Workspace pane

Step 2: Use the Selection Tools

Metashape offers several selection tools to isolate noisy areas:

  • Rectangle Selection (R): for box-like areas
  • Free-Form Selection (F): for organic shapes
  • Circle Selection (C): for targeted spots
  • Connected Component Tool: isolate floating points

Step 3: Delete Unwanted Points

After selecting noisy data points, press Delete or go to Edit > Delete Selection. Be cautious not to remove valid points near edges or thin structures.

Step 4: Use the Classify Ground Tool (Optional)

For terrain models, use the Tools > Dense Cloud > Classify Ground Points feature. This separates ground and non-ground points (trees, buildings). You can then delete vegetation points for a cleaner DEM or DTM.

Step 5: Apply the “Filter by Confidence” Tool

Metashape assigns a confidence level (0–255) to each point based on photo coverage and quality. Go to:

  • Tools > Dense Cloud > Filter by Confidence
  • Set a threshold (e.g., 100 or 150)
  • Remove points below this value

This automatic method helps remove poorly reconstructed areas.

Step 6: Save and Export

After cleaning your cloud:

  • Right-click the Dense Cloud layer
  • Select Export Dense Cloud
  • Choose formats like LAS, PLY, XYZ, E57

Best Practices

  • Use high-resolution, overlap-rich images (80/70%)
  • Avoid windy or rainy conditions
  • Capture oblique angles to fill in gaps
  • Review each region before bulk deletion

When to Use External Tools

If your project requires even finer control, you can export the point cloud to external tools like:

  • CloudCompare: For segmentation, noise removal, and comparison
  • Trimble RealWorks: For infrastructure and construction scans
  • Blender: For visual cleanup and mesh correction

Conclusion

Cleaning noisy point clouds in Agisoft Metashape is a critical step in professional 3D workflows. Whether you’re modeling archaeological sites, generating DSMs, or building BIM-ready datasets, a well-refined dense cloud ensures accuracy, visual clarity, and compatibility. By following this guide, you’ll gain better control over your photogrammetric outputs and avoid downstream errors.