Datasets

On-Demand Processing

Registry automatically generates optimized formats for visualization:

Source	Generated Format	Description
GeoTIFF	COG (Cloud Optimized GeoTIFF)	Efficient streaming for large orthophotos
Point Clouds	EPT/COPC	Streaming format via Potree viewer
3D Models	NXS (Nexus)	Progressive streaming for 3D meshes
Images	Thumbnails, WebP tiles	Fast previews and map tiles

Processing happens in background via Hangfire jobs. Monitor progress at /hangfire.

Panorama Viewer

Registry includes a built-in 360° panorama viewer for immersive visualization:

• Automatic detection: Images with aspect ratio ≥ 2:1 are classified as panoramas
• GPS support: GeoPanoramas include location data displayed on maps
• Interactive controls: Pan, zoom, and fullscreen viewing

To upload panoramas, simply add them to a dataset like any other image. Registry will automatically detect and render them with the panorama viewer.

Chunked Uploads

For large files, Registry supports chunked uploads:

• Files are split into manageable chunks for reliable upload
• Automatic resume on connection failures
• No file size limits (configurable via MaxRequestBodySize)

The DroneDB CLI (ddb push) handles chunked uploads automatically.

User Management

Registry includes a comprehensive user management system with role-based access control.

Full Documentation

For complete details on user management including API endpoints, roles, organizations, storage quotas, and authentication, see the dedicated User Management Guide.

Key Features:

• User account creation and administration
• Role-based access control with custom roles
• Organization membership management
• Per-user storage quotas
• JWT authentication with external provider support

Dataset Visibility

Datasets support three visibility levels:

Level	Description
Private	Only the owner and admins can access
Unlisted	Accessible with direct link, not listed publicly
Public	Visible to everyone, included in STAC catalog

Change visibility using:

• Web UI: Dataset settings
• CLI: ddb chattr +public or ddb chattr -public

Dataset Deletion

When you delete a dataset, Registry uses a deferred deletion approach for reliability:

1. Immediate: The dataset is removed from the database instantly, so it disappears from the UI
2. Background cleanup: A Hangfire job handles:
   • Cancelling any active build jobs (tiles, thumbnails, 3D conversions)
   • Removing job tracking entries
   • Deleting the filesystem folder

This approach ensures that:

• Users get immediate feedback when deleting datasets
• Active build processes don't block deletion
• Locked files (e.g., during 3D model conversion) don't cause errors

If the background cleanup fails (e.g., files still locked), a daily recurring job (cleanup-orphaned-datasets) will automatically clean up any orphaned folders.

info

You can monitor cleanup jobs in the Hangfire dashboard at /hangfire on your Registry instance.