I am trying to process a farm of 3500 hectares which I have taken 9 flights of my aerial drone. How to process them all in a unique Pix4d Project.
I have tried to calibrate which one before and them open a new Project by 9 existing projects. Have I done it correct? if yes, why it is not working?
Kind Regards,
Luiz
As we have further investigated the case through direct support. We will summarize our answer here, so that if someone has the same issue he has some pointers on how to improve the project.
The recommended workflow for large projects is described in this article from our Knowledge Base: https://support.pix4d.com/hc/en-us/articles/202558579
There were three main issues in the project:
The difference between the initial and optimized internal camera parameters was about 14%, this is visible in the summary of the quality report:
This means that the optimized camera model is significantly different from the initial parameters, which could mean that the reconstruction does not correspond to what it should (as it uses parameters that are too different). We recommend that this value is smaller than 5%.
In this case, the issue was already present in the subprojects. Always make sure that the subprojects are well reconstructed before merging. In order to fix this, we would:
a) We would double check if the camera parameters (e.g. focal length) were correctly initialized. The camera model parameters can be edited as described here: https://support.pix4d.com/hc/en-us/articles/202560169#label2
b) If the camera parameters were correctly initialized, it could be that there were several possible optimizations of the camera parameters. In this case we would use the “All Prior” option, which tells the software to select the optimized value closer to the initial parameters:
After these changes, step 1. Initial Processing of the subprojects should be run again. Then, the subprojects can be merged once more.
In the merged project, there were 2 blocks.
This means that at some point in the reconstruction, the software was not able to find enough similarities between two adjacent areas, so it optimized them separately.
In order to fix this, we would add 3 to 5 Manual Tie Points (MTPs) between the two blocks, marking images from both blocks in the same MTP:
Note, we recommend to add MTPs in each subproject, on areas that are overlapping with the adjacent subprojects, so that the merging is improved:
There was an issue that the orthomosaic was too large to be used in AutoCAD. The size can be reduced by reducing the resolution or by using a Processing Area. More in this article: https://support.pix4d.com/hc/en-us/articles/202559989
