When information about a site is not available, it is interesting to make a survey by drone to realize a photogrammetry. The photogrammetry will allow to build a 3D model and to extract the necessary information to understand the site.
The preparation of the drone flight is essential. The flight must be able to cover the entire site from different angles. Otherwise, there would be deformations or missing information.
The example presented here presents the workflow for creating a 3D model from an aerial film shot with a drone and the processing of the data for further work.
Here, the drone was forced to work in sections. Here are the different views. The work to come will be to assemble the 3D file fragments.
We will work from a drone survey done at the location Batterie de la Reine situated here : -20.375756949812796, 57.728543193103164
extraction of images
For each segment, we use the free software Free Video to JPG Converter (https://www.dvdvideosoft.com/products/dvd/Free-Video-to-JPG-Converter.htm) to convert the movie to jpg images.
The choice of the number of images depends on the length of the sequence and the movement of the drone. It is therefore necessary to make tests according to what has been filmed. The idea is to have images that follow each other in a fluid way, but still, not to have too many so as not to penalize the calculations.
In short, you have to test.
The extraction creates a large collection of images. Beware, photogrammetry consumes a lot of storage space, it must be planned in advance.
We will use Metashape, Agisoft’s flagship software, one of the best for photogrammetry. It can be loaded here. https://www.agisoft.com/
The workflow is quite simple, just let yourself be guided by the menus.
Open Workflow Menu
Then compute Align Photos
Operate as followed
You get a light CloudPoint. To view the camera path activite Show Camera button.
Go back to workflow and activate Build Dense Cloud
To visualize the result Click on Dense Cloud button
Then back to Workflow and choose Build Mesh
To visualize the result Click on Model Shaded button
Last step, go back to Workflow and click on Build Texture
To visualize the result long Click on Model Shaded button to activate Model Textured button.
Calculating other parts
In the Workspace, add new chuck, and then restart the process with a new set of photos
To assemble, we need to choose common geometry between the two Chunks.
We have this in the first one
And that in the second chunk
We then add markers on both common geometry on each chunk, with the same name and same order. Right Click, add marker.
When done go to workflow, Align Chunk, choose Marker Based
It results a Merged Chunk in the WorkSpace.
The two parts are combined
Before, clean up what is not too good by using selections tools and delete to erase unwanted polygones.
To have a Merged model go to WorkFlow Merge Chunks
Then, remesh and recalculate the texture.
We can do a bit more of cleaning
The easiest way is to select what needs to be kept and then invert selection
Scaling the model
Remove all markers (select and right click remove markers) and add new one on the known geometry.
We add 3 markers which will be working as orthonormal reference frame.
Then we change the values of the points to input the ones we want
If not it won’t work
Exploring in VR and AR
The model is ready to be exported. Sketchfab