TUTORIAL HOME

Getting started with 3DF Zephyr

 

Welcome to the 3DF Zephyr tutorial series. In this tutorial, you will learn the basic steps for turning pictures into 3D models with 3DF Zephyr.
IMPORTANT NOTE: The images in the following tutorial describe the full version of 3DF Zephyr, showcasing all its features. Please note that 3DF Zephyr Free and 3DF Zephyr Lite follow the same workflow for 3D reconstruction, but not all options may be available. You can consult the 3DF Zephyr Feature Comparison page to view all available features in each version.

 

 

 

Summary

 

 

 

xxxxxxxxxxxxxxxxxx

 

 

  • Introduction

Return to summary
Creating 3D models from pictures in Zephyr requires a dataset of images taken using photogrammetry techniques. You can use your own pictures by following this tutorial. Please refer to these base guidelines to learn the best practices for acquiring a photogrammetry dataset. Note that the image acquisition phase is crucial for successful photogrammetry reconstruction.

NOTE: before starting a photogrammetry reconstruction in 3DF Zephyr, it is important to provide a reliable dataset. Avoid using blurred images and datasets with no overlapping pictures, as these can lead to poor quality reconstructions. Please never crop, cut, or edit the pictures in Photoshop or other image editors.

 

If don’t have your pictures yet, you can use our sample dataset below. Please download the Dataset – Cherub zip file and extract it on your computer.

This dataset is composed by 65 photos, if you’re using 3DF Zephyr Free, which is limited to 50 photos, simply select the first 50 photos only.

download datasetDownload Dataset – Cherub (531MB)
Download – Cherub .ZEP file (340MB)

 

If it hasn’t been done already, the following links are available for downloading and installing the suitable version of 3DF Zephyr to start the tutorial:

3DF Zephyr3DF Zephyr Lite3DF Zephyr Free

Download – 3DF Zephyr

Download – 3DF Zephyr Lite

Download – 3DF Zephyr Free

  • Step 1 – Creating a new project

Return to summary

  • Starting the new project wizard

Return to summary
To create a new project, click on “Workflow” > “New Project” (1). The Project Wizard (2) will guide through the picture import and camera orientation phase setup. It’s possible to select the required 3D reconstruction phase(s) in addition to the camera orientation phase by enabling the corresponding icons in the Computation steps (3) graph.

  • SPARSE: This option will orient the images and generate a Sparse point cloud.
  • MODEL: This option will create the Sparse point cloud, Dense point cloud, and Mesh.
  • TEXTURE: This option will include all the previous 3D products and generate a Mesh with texture.

 

For example, enabling the “MODEL” icon will result in the software completing the Sparse point cloud, Dense point cloud, and Mesh generation before stopping and allowing selection of the Textured mesh generation.

 

Depending on the version of Zephyr used, the Additional option (4) section can include the following check-boxes: Import of image masks, Define constraints (2D and 3D coordinates), and Setup of multi-camera groups.

If the additional options are not needed, simply click the “Next” button to proceed through the wizard.

Note: only in 3DF Zephyr Full version, it’s possible to decide in the final phase whether to generate a Mesh with texture or an Orthophoto on the Z axis. For the advanced orthophoto workflow, follow this link: Section A: Generating orthophoto via new project wizard.

 

If only the Sparse Point Cloud creation phase (“SPARSE” in the graph) is chosen, it is still possible to start the other phases subsequently from the Workflow menu selectively. Here is a brief description:

  • 3D Model Generation: Creates a dense point cloud and extracts the mesh in a single step.
    Requirement: Sparse point cloud
  • Texture Mesh Generation: Creates a mesh with texture.Requirement: Mesh

 

From Workflow > Advanced, the following options will be available:

  • Dense Point Cloud Generation: Creates a dense cloud.
    Requirement: Sparse point cloud
  • Mesh Extraction: Creates a mesh with RGB colors.
    Requirement: Dense point cloud

 

  • Selecting the images

Return to summary

The next window will present the Photo selection page, where you can add images to be processed by Zephyr.

Click on the “Add directory” (5) button, browse to the directory where your dataset is located, and then press the “Select folder” button. If there are subdirectories in the folder, Zephyr will prompt you with a pop-up window to include them in the wizard by clicking a “YES” or “NO” button. You can also drag and drop images or folders directly into the Photo selection space.

Alternatively, you can click the Add images (6) button, browse to the directory where your dataset is located, select all the images you previously extracted, and then click the “Open” button. If needed, you can select one or all images (ctrl+a) and click the “Remove” (7) button to exclude them from the workflow.

Usually, most datasets are created using .JPG or .PNG files, but Zephyr can also use many other image formats, such as raw files (NEF, CR2, ARW2, etc.).

When all the pictures are selected, press the “Next” button to proceed to the following wizard page.

 

  • Camera calibration phase overview

Return to summary
Once the data to be processed is selected, the Camera Calibration page will appear. Here, it is possible to Add a New Calibration or Modify the Calibration that 3DF Zephyr has automatically selected.
The interface will display a list of the imported pictures along with their calibration summary, which 3DF Zephyr and its Structure from Motion algorithm will specifically utilize. When connected to the internet, 3DF Zephyr can download pre-calibrated camera parameters to speed-up the processing. However, the software is fully auto-calibrated, allowing it to function offline, and in most cases, all the parameters can be left as default.

 

Click the “Next” button to proceed to Step 2: Configure the reconstruction settings.

  • Step 2: Configure the reconstruction settings

Return to summary

In this step of the tutorial, the settings pages of 3DF Zephyr will be illustrated. Depending on the 3D reconstruction phase(s) chosen in step 1 of the tutorial, the respective setting pages will consistently appear.

  • Camera orientation and sparse point cloud generation

Return to summary
The first phase of 3D reconstruction is called Structure from Motion and computes the position and orientation of each photo loaded in Zephyr. This step results in a Sparse point cloud.
The software analyzes each image, identifies its features, and compares each image with (usually a subset of) the other pictures. This process is used to set the cameras in the correct position.

Users must define the Category depending on the scenario/subject they have captured and select the Preset according to the accuracy and quality level they want to achieve on the 3D model(s).
Click the “Next” button to proced.

For this tutorial, the following settings will be used:

Category

    • :

General
Preset

    :Default settings.

 

IMAGE

 

The Advanced settings mode allows to “tweak” the reconstruction engine process. It’s possible to switch from Preset Mode to the Advanced settings by clicking the drop-down menu in the upper right corner of the window.

Advanced settings overview (click to open)
The Advanced settings are explained in detail below:
 

MATCHING

Keypoints Sensitivity: “keypoints” refers to the matched pixels in the pictures. This parameter determines the maximum number of keypoints that are to be extracted for each image. Please note that a substantial number of keypoints does not lead to a better result since the more elements involved, the higher the ambiguity during the matching phase.
 

  • Low: extracts up to 5,000 keypoints per image
  • Medium: extracts up to 7,500 keypoints per image
  • High: extracts up to 10,000 keypoints per image
  • Very High: extracts up to 15,000 keypoints per image

 
Keypoints Sensitivity can be:

Decrease
Increase
  • the outcome is required quickly;
  • the distance between the cameras and the subject/scene does not vary;
  • pictures have enough (at least 60%) and high (70% – 80%) overlap;
  • cameras are not oriented properly;
  • pictures are too close to or too far from the subject/scene;
  • pictures have low overlap;
  • the subject/scene has a uniform or textureless surface (e.g., a white wall);

 
 

Matching Type: controls which optimizations (if any) should be used when computing matches between images. The Accurate option will take longer to complete camera orientation;

 
 

Matching Stage Depth: controls how many keypoint matches are performed on the images. Usually, the more is the better, however, this comes at a computational cost.

Matching Stage Depth can be:

Decrease
Increase
  • the outcome is required quickly;
  • the distance between the cameras and the subject/scene does not vary;
  • pictures have enough (at least 60%) and high (70% – 80%) overlap;
  • cameras are not oriented properly;
  • pictures are too close to or too far from the subject/scene;
  • pictures have low overlap;
  • the subject/scene has a uniform or textureless surface (e.g., a white wall);

 
 

Photo Ordering: this parameter tells 3DF Zephyr which capture sequence has been leveraged onsite, thus speeding up the near-image detection. By using the Sequential option, you are telling 3DF Zephyr the pictures have been taken sequentially (photo 1 – photo 2 – photo 2 – photo 3 – etc.), while the “Circular” option hints at a closed loop sequence. If you don’t know the photo order, pick the Unordered option. The last Approximate grid option is designed for nadiral aerial datasets.

 
 

RECONSTRUCTION

 

Tie-points quantity: this value affects the number of points to be considered during the Bundle Adjustment optimization. A higher option can lead to better results and a higher computational cost.
 
 
Reconstruction Engine: 3DF Samantha algorithm supports both Incremental and Global reconstruction pipelines.

    – The Global pipeline tries to solve the orientation problem globally, which means that it’s usually faster than the incremental one and potentially less prone to drifting error.

    – The Incremental pipeline on the other hand, albeit slower, tends to include more cameras in the final solution. In a nutshell, you are supposed to use the Global pipeline if you are processing a huge dataset (more than 1000 pictures).

 
 
Bounding box method: it defines the computation method of the bounding box. It’s recommended to keep the default option “discard outliers”, namely the far points that are not relevant to the 3D reconstruction.