Support Website Contact Support Blog

Scattered Image Processing

Hi everyone! I am processing some imagery from a field site and both missions were executed with Pix4D Capture using a DJI Phantom 4 pro v2. The parameters were all the same and were only taken maybe a day apart. I don’t know or understand why I am getting such different imagery composition between the two missions. I’ve attached the photos of the imagery. The image dunlen04 is scattered everywhere, and the camera rays, images are no where to be seen, and the map of the double grid looks great on the flight plan summary before processing? Dunllc01 is more what I am looking for as I am trying to extract microtopography from the imagery.

Anyways, I am having this scattering issue on more than just this one site, and I was wondering if there is anything I can do to fix it? Or if anyone has had this issue before?

Thanks in advance!

Callie

note I could only attach one photo, so I attached the one with the image all scattered and messed up

here is the photo of the one that worked (I think, very new to this software as I am doing a graduate project with this process and seeing if it will work for habitat classification!)

Hi Callie,

Can you please share with us the quality reports of the 2 projects in .pdf format (the scattered and the successful one)?

You will find them under the path: …\project_name\1_initial\report\project_name_report.pdf

Best,

Hi Nikoleta! Thank you so much for your help!

Attached are the quality reports, again the one titled DUNLEN04 is one for the scattered images!

DUNLEN04_report.pdf (1.0 MB)
DUNLLC01_report.pdf (2.2 MB)

Hello Callie,
I reviewed your two projects and immediately noticed the two datasets were not set with the same parameters. The good project was processed with 89 images and the problematic one had 62 images. So there was 30% less data for the software to work with. In many instances this may not be a problem but your project area appears very homogenous and uniform. In this instance be cognizant of your image overlap and consider increasing it to 80-90%.

yeah! I noticed that too with this specific project (i didnt’ check the amount of images, and now realized I was missing some).

I’m going to attach another one, called DUNLRH04 and this one has a similar amount of images with 80/90% overlap and they are scattered again. It was however, completed with a different drone and has images with red boxes instead of blue during processing? I am not sure what that may mean.

I have a third one as well, that has 80-90% overlap that I will also attach to this thread. DUNLRH04_report.pdf (1.5 MB)

Hi Callie,
The red boxes mean that the images were uncalibrated or were rejected by the software. There are several reasons why this could happen but typically it has to do insufficient keypoints. It appears that you are using a DJI X5 camera. Since this is a trusted camera I would suggest setting your Internal Parameters Optimization to All Prior. This will force the optimal internal parameters to be close to the initial values. Basically, the software will trust this camera. Also set the calibration method to Alternative.

Give this a shot.

ahh! thats so helpful! I will run a couple sites with those settings! I was also considering changing the Keypoints Image Scale from 1/2 to 1/4 or 1/8. Do you think that would help as well? I am trying to get a good idea about vegetation/topography in a (like you said) fairly homogenous landscape of wetlands!

Thank you so much for your help!

Hi Callie,
Reducing the Keypoint Image Scale can be useful when trying to model vegetation. The problem with vegetation is that it has complex geometry that tends to move due to wind. So reducing the keypoint image scale can help PIX4Dmapper reduce noise and generate better keypoints over vegetation.