Create DSM and merge several subprojects

Hello, good morning, I am working with PIX4D, but I have doubts about how to generate the MDT and perform subproject mergers.

For the generation of the MDT that it has, PIX4D almost always does not do well the selection of the objects that are not proper of the earth, as it is the vegetation, the houses and other things, then that is where the selection has been done manually to identify what things are specific to the earth and what are not. My case is that I am taking the photographs where most of the objects are vegetation, so my question is how high will it take me in those areas when I remove the vegetation (for when they want to make the contour lines) and if it is possible to do a faster selection for objects that are not part of the terrain, since it is very slow when the area is very large.

As for the merging of the subprojects, I saw a PIX4D tutorial where you have to execute process number 1 in each subproject and then merge them and process 2 and 3. My question is if you can perform process 1 and 2 in each subproject and then merge them and do the process 3. (The total area to be built is approximately 46 km2)

I hope you can help me.

Thank you!

Hola Luis,

Digital surface model (DTM)
To generate a digital surface model (DTM), only the points in the Ground and Road Surface groups will be retained. As a consequence, you need an accurate point cloud classification. For this, indeed, you generally need to “correct” the classified point cloud by editing it, i.e. re-assigning points to the correct groups (Ground, Road Surface, High Vegetation, etc.). Note that, for now, there is no other way to edit the point cloud than manually.
In the areas where there are no points belonging to either Ground or Road Surface, the DTM will be interpolated according to the elevation of the surrounding points.

With photogrammetry, you can only render the content of the images. As a consequence, if no ground is visible in the images because they are filled with vegetation, no ground will be reconstructed and there will be nothing with which to generate the DTM.

Merging projects
For now, what you suggest is not possible. It is only possible to merge the results of step 1 when merging projects, not of step 2.
What would merging projects after step 2 achieve/allow you to do that the current procedure does not? Could you give me more details so that I can better understand your needs?

Thank you very much for replying to Rhéa Garratt. : D

Regarding the fusion of projects, my concern is that an area of approximately 46 km2 will be built (it is a very large area). We are testing 667 images corresponding to an area of 1.12 km2, so calculating the rest the images would be little more than 27,000 images. We will work with 2 computers (Core i9 X-series processor and 32 GB of RAM). The intention is for each computer to process the images taken on each flight to go after manually selecting the objects for the MDT and then joining them together in a single project. But I think that in each flight will only be able to process step 1 and I will have to wait until I finish building the 46km2 that is planned and then process all the images along with steps 2 and 3 so that the ortomosaic of all that area show in a single image. My question now is whether the 32 GB of RAM will be enough to process all that amount of images.

And there will be no way that I can by are manual classification of the objects (for the MDT) for each flight made? is that waiting to be processed all the area that is intended to be lifted will be very complicated.

Hi,

You are welcome :slight_smile:

You can find our minimum and recommended system requirements here. To process over 10’000 images, you should definitely have a very good computer with more RAM than 32 GB. Users and one of my colleagues raise some good points, regarding hardware, processing time and use of the outputs in third-party software in the following community topic: Experience with large datasets?. I recommend you read the thread and comment on it to make it active again.

If what you are interested in are the DTM and the orthomosaic, you could consider generating these outputs for each sub-project, then merging them in a third-party software (for example QGIS). To make sure the edges of the maps (DTM and orthomosaic) for each subproject are contiguous (no overlap and no gaps), you should use a processing area in each subproject. I recommend you create contiguous areas in a GIS software (for example QGIS), export them in the shapefile format and import them in your subprojects.

The workflow for each subproject would be the following:

  1. Process step 1.
  2. Add a processing area.
  3. Process step 2.
  4. Edit the point cloud.
  5. Generate the DTM and/or the orthomosaic.

Then, you could merge the tiles in a third-party software (for example QGIS).

I hope this helps!

thank you very much
It is what I also found, I will do then. I am using Global Mapper and ArcGis in both I have been able to make the union of the ortomosaics, the MDT and MDS, I will continue testing, I thank you very much for your attention. :slight_smile:

Good to know it already worked well for you. Always happy to help!

And do not hesitate to continue sharing your experience here on the community.