I’m working on a big field project.
I cut the project into two sub-projects that are good respectively, including rendering the point cloud.
The merging of the two projects using manual points is good, but the next step is less good. The point cloud associates images taken at different times with shading problems or image quality … this makes cleaning tedious. Point cloud stand alone for each project is better. Is there a solution for the point cloud of the complete project to be built only on a chosen series of images?
By the way, when cleaning (sky, vegetation …) can we clean by choosing a color (blue for example)?
You might want to combine/clean the two projects in something besides Pix4D, which isn’t always the best tool to use
Also be sure to Reoptimize after merging the two projects, it isn’t required and has to be done manually but I have never had a good merge without it.
For large areas, I would suggest to fly small blocks at a time, where each block has overlap with the next. This improves the similarity of images in each block and thereby reduces the issue you mention. If necessary, the merging procedure can be used to combine the flights on the overlapping areas. The problem would then be limited to the overlapping area only.
It would be very helpful if you could provide screenshots for a comparison of the individual projects and the merged project. A quality report of your project would help as well to give more specific advice.
If there are issues in the quality of your images, there might be issues in step 1 already. If there is an issue with noise in the densified point cloud, you could increase the minimum number of matches necessary for the densification, which should reduce the noise:
I suspect that there are other issues in your project and I would heavily recommend to make sure that these are fixed before. However, to answer specifically to your question: Is there a solution for the point cloud of the complete project to be built only on a chosen series of images?
It should be possible to create a densified point cloud based on a subset of the images. First, you would have to split the images in more than one group in the Image Properties Editor:
Then, you can select based on which group you would like to create the densified point cloud in the processing options:
Then, import it to generate the DSM:
This function is usually used for multispectral imagery. This article says more about it.
Last but not least, points cannot be cleaned based on color. However, to remove sky from the project I would suggest the Carve tool, see here.
thank you very much for your very detailed and constructive help.
I will try your solutions and give you a return.