- Asset-centric (legacy): assets are asset-centric, and contextualization approvals are typically stored via the Annotations API (for 360° images) or classic 3D APIs.
- Hybrid: you can contextualize both asset-centric assets and core data model assets. Approvals are stored using 3D asset mappings (CAD) and the Annotations API (360° images and point clouds).
- Data modeling: assets are core data model instances (for example, CogniteAsset). Contextualization annotations are stored in Cognite data modeling, but you create them through the 3D contextualization APIs. For point clouds, 3D jobs can generate contextualization suggestions.
Suggestions vs. approvals
It helps to separate contextualization into two stages:- Suggestions: automated output you review (for example, point cloud volumes or detected text regions in 360° images).
- Approvals: the links you create to store the contextualization (for example, “this 360° text region refers to this asset”).
Which API to use
The table below summarizes the main API entry points for 3D contextualization.- Asset-centric and hybrid
- Data modeling
| Data type | Suggestions | Contextualization | Procedure |
|---|---|---|---|
| CAD | Filter 3D nodes / List 3D nodes | Create 3D asset mappings | Contextualize CAD models |
| Point clouds | Create 3D revisions | Create annotations | Contextualize point clouds |
| 360° images | Vision extract (deprecated) or your own tooling | Create annotations | Contextualize 360° images |
For 360° text suggestions in asset-centric and hybrid projects, the Vision API is deprecated. We recommend migrating to a data modeling project.