🤗4.1 Cluster Computing

Get3D Mapper supports multiple computer Cluster Computing to process project with large number of images. Here we introduce you how to batch management of Engine nodes for cluster computing.

Note:

(1) Before performing cluster computing on multiple computers, please make sure that each computer is connected to same LAN and can connect each other. Each machine has read and write permission to other computers.

(2) Get3D Mapper software contains two modules: Get3D Mapper (UI for user to operate) and Get3D Engine runs at background (algorithms for automatic modeling).

1 Introduce

Get3D Mapper supports multiple computer Cluster Computing to process project with large number of images.

Here, we introduce you how to batch management of Engine nodes for cluster computing.

2 Download & Installation

First, download and install Get3D Mapper on each computer.

The Get3D Mapper installation process will automatically accompany the installation of Get3D Engine.

3 Login Account &Obtain License

After finishing the installation,

(1) open the Get3D Mapper software on each of the 10 computers,

(2) log in to your account and activate the licence to each computer.

The tutorial of how to obtain license:

https://docs.get3d.ai/user-help-center/get3d-mapper/how-to-use-it/3.1-start-tutorial#id-2.-license

4 Start Engine

Start the engine on all 10 computers:

click Settings-Local Engine Settings in the menu bar,

then click Start Engine to access the Engine UI interface

5 Create Project

After obtaining authorization and starting the engine on all 10 computers.

Select a computer as the master computer and create project.

Please set the project path to a network path that can be accessed by other computers. Subsequent operations can be completed on the project of this master computer.

Note:

All operations except Collaborative Mark Ground Control Points can be performed only on the main project of the master computer:

  • creation of the project;

  • import of the images;

  • positioning of the images;

  • submission of the Relative Orientation (Aerial Triangulation on the free network);

  • submission of the Absolute Orientation (Aerial Triangulation on the control network);

  • Submission of Reconstruction (generation of the 3D mesh model).

Similarly, mobilizing the computational power of each engine node, managing the start or pause of the engine nodes, etc., can be done from a single engine interface.

The tutorial of how to create project:

https://docs.get3d.ai/user-help-center/get3d-mapper/how-to-use-it/3.1-start-tutorial#id-3-create-project-and-open-project

6 Setting Engine Path

After creating the project, set the engine path. Open the Setup Engine window in the menu bar Settings.

Set both the Task Directory Settings and the Local Engine Setting to the same network path.

Note:

There are four paths in the cluster calculation that need to be network paths:

① 【Create & Open Project】 Project Path

② 【Engine Settings】 Task Directory Path (② and ③ are consistent)

③ 【Engine Settings】 Local Engine Path (② and ③ are consistent)

④ 【Import & Read Image】 Image Path

The tutorial of how to setting engine path:

https://docs.get3d.ai/user-help-center/get3d-mapper/how-to-use-it/3.1-start-tutorial#id-6-engine-setting

7 Import Image & Positioning

The tutorial of how to import images and positioning

https://docs.get3d.ai/user-help-center/get3d-mapper/how-to-use-it/3.1-start-tutorial#id-4-import-image

8 Submit Task & Distribute Engine Nodes

Task submission:

After completing the image import and positioning, and correctly filling in the Camera Information (Sensor Parameters and Focal Length), click Submit Aerial Triangulation.

Distribute Engine Nodes:

(1) Open the engine interface. In the Engine Information column, you can see the current state of the Engine (Waiting, Pause, Start), the version of the Engine, and the address of the Task Path and so on.

(2)Select all Engines, right-click on Task Path and select Modify Task Path. Setting the Task Path of all the 10 Engines to the task path set when the main project submits the task.

(3)Start Engine: Right click to Start, then all the Engine of ten computers start cluster computing for the task.

Last updated