Fly double grid mission (images are geolocated), perform Step 1
Fly circular mission (images without geolocation), perform Step 1
Enter 8 same MTPs in each project
Merge projects
Perform Step 1,2
Edit density point cloud and classify false points as disabled
Run Step 2. Visual inspection of resulted Densified point cloud show good quality of data
Run Step 3. Visual inspection of resulted orthomosaic shows poor quality – I cannot see the helicopter tail.
If I displays the density point cloud with minimum point size from the Top and if I choose rayCloud->Perspective/Orthographic – I can visualize the helicopter and landing pad much better that generated orthomosaic looks like.
What you can recommend me in this situation?
P.S. I use High Resolution Settings (16384x16384) for Step 2 and I use 1x GSD resolution for DSM and Orthomosaic.
Unfortunately, I do not have any opportunity to fly over this place again. But I have processed orthomosaic separately for double grid and circular missions. Attached images that looks much better than orthomosaic for merged projects.
I thought that the better the quality of density point cloud the better the quality of DSM and the better quality of resulted orthomosaic, that is why I decide to merge this two project together.
As a result I can see very detailed point cloud in PIX4D rayCloud view window. The problem is that I still don’t understand exactly why quality of orthomosaic for merged project is worse than the orthomosaic produced from separate projects.
If for example I made any mistake while merging (position of MTPs in each project doesn’t correspond to each other perfectly), why I can see very detailed point cloud in rayCloud window? And what is the best way to merge two projects as accurately as possible and what is the best way to evaluate the resulted alignment accuracy?
Sorry for the long delay. As partially I accidently deleted some of initial results, I decided to rebuild all projects again. Now results seems to be very good for me: both ortho mosaic (attached final image) and density point cloud.
But there is difference between projects (initial and final):
In initial project: I took only 1100 images for circular mission (about 400 degrees) (trimmed version) and after merging meshes I Edit Densified Point Cloud and mark noise points as disabled;
In final project: I took 1593 images for circular mission (about 720 degrees)(full version) and after merging I did not Edit Densified Point Cloud.
I put all reports in OneDrive.
Based of this result, if possible, i ask to give me some reference on “Best Practice” on projects merging (how many MTP should i create, do i need to mark them on every frame, etc) and on editing point cloud process (what is the best way to evaluate the results?).
In both projects, you used _Multiscale _processing option. I would highly recommend unselecting it as in your dataset this option creates lots of noise which at the later step you have to eliminate.
Also, in the initial project, the temple you used was 3Dmodel, which is correct when you want to model an object. However, in the final project, the chosen template is 3DMaps. Could you tell me what the outputs of your interest are? Do you want to model this helicopter or create only an orthomosaic?
We can also decrease high error in Camera optimisation by choosing All Prior in Internal Parameters Optimization as the relative difference between initial and optimised internal camera parameters is above 10% each time.
Regarding “Based of this result, if possible, i ask to give me some reference on “Best Practice” on projects merging (how many MTP should i create, do i need to mark them on every frame, etc) and on editing point cloud process (what is the best way to evaluate the results?).”
I have to say it is difficult to give a general rule as it depends on a dataset. However:
How many MTP should I create?
Enough to be able to merge the project accurately and if the point clouds of subprojects do not fit correctly and some “point levels” are created - in those areas you should add more MTPs to eliminate this problem.
Do I need to mark them on every frame?
To have accurate results, yes. The more pictures that are tagged, the lower the error. Please take advantage of Automatic Marking once you are confident that marks are OK.
What is the best way to evaluate the results?
I’m sorry. I don’t understand this question. Could you explain it to me? Thanks!
My last recommendation would be to update your NVIDIA graphic card to the newest version. To do so, please use this link.
Hi Beata,
Many thanks for your detailed answer!
I will reprocess all projects again taking into account your advice.
Regarding project types: initial intention was to create orthomosaic as acurate as possible. But the opportunity to measure any dimensions of any objects in virtual 3d space also very important to me. Based on that, what types of project should I use (if I want to get both: 3D model and orthomosaic)?
Regarding my last question “What is the best way to evaluate the results?” – I mean how I can understand that the projects are merged together with sufficient quality and there is no need to improve anything? If I did not see noise that is the sign of good merge quality for example?
For orthomosaic: Use the double grid dataset with 128 images. Select 3D Maps template but apply some changes in Step 2, which are; unable Multiscale option as I mentioned before, decrease the minimum number of matches to 2. In case 2D Keypoint Matches are still quite weak,
For 3D Model: Use your oblique dataset with 1200 pictures, select 3D Model template, apply All prior in Step 1, Calibration tab, decrease the minimum number of matches to 2 in Step 2
Once those two projects have the right parameters/accurancy, you can merge projects.
For your question “How I can understand that the projects are merged together with sufficient quality and there is no need to improve anything?” our video academy How to ensure quality and accuracy in Pix4Dmapper, is an answer :-)
These cookies are necessary for the website to function and cannot be switched off in our systems.
They are usually only set in response to actions made by you which amount to a request for services, such as setting your privacy preferences,
logging in, or filling in forms. These cookies do not store any personally identifiable information.
These cookies allow us to count visits and traffic sources so we can measure and improve the performance of our site.
They help us to know which pages are the most and least popular and see how visitors move around the site.
All information these cookies collect is aggregated and therefore anonymous.
If you do not allow these cookies we will not know when you have visited our site, and will not be able to monitor its performance.
These cookies may be set through our site by our advertising partner (Google).
They may be used by Google to build a profile of your interests and show you relevant adverts on other sites.
They do not directly store personal information but are based on uniquely identifying your browser and internet device.
If you do not allow these cookies, you will experience less targeted advertising.
